Upload folder using huggingface_hub

.gitignore

@@ -0,0 +1,2 @@
+node_modules
+.claude/

README.md

@@ -1,12 +1,73 @@
----
-title: Inference Metrics
-emoji: 🏃
-colorFrom: gray
-colorTo: purple
-sdk: gradio
-sdk_version: 5.39.0
-app_file: app.py
-pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# LLM Pricing
+
+A tool to fetch and compare LLM pricing and capabilities across multiple providers.
+
+## Data Sources
+
+This tool uses two primary data sources:
+1. **HuggingFace Router API** (https://router.huggingface.co/v1/models) - Primary source for model pricing, context length, and capability flags
+2. **Provider-specific APIs** - Fallback source for additional metadata and capabilities
+
+The HuggingFace Router API now provides comprehensive data including:
+- Pricing (input/output costs per million tokens)
+- Context length
+- supports_tools flag
+- supports_structured_output flag
+- Provider status
+
+When data is available from both sources, the HuggingFace Router data takes priority.
+
+## Installation
+
+```bash
+bun install
+```
+
+## Usage
+
+```bash
+# Fetch all models and enrich with provider data
+bun run get-metrics.ts
+
+# Skip specific providers
+bun run get-metrics.ts --skip-providers novita featherless
+
+# Test performance for models (requires HF_TOKEN)
+HF_TOKEN=your_token bun run get-metrics.ts --test-performance
+
+# Test specific number of models
+HF_TOKEN=your_token bun run get-metrics.ts --test-performance --test-limit 10
+```
+
+## Supported Providers
+
+- **novita** - Full API support
+- **sambanova** - Full API support  
+- **groq** - Full API support
+- **featherless** - Full API support
+- **together** - Full API support
+- **cohere** - Full API support
+- **fireworks** - Full API support
+- **nebius** - HF Router data only
+- **hyperbolic** - HF Router data only
+- **cerebras** - HF Router data only
+- **nscale** - HF Router data only
+
+## Output Files
+
+- `enriched_models.json` - Complete enriched model data
+- `provider_models_raw.json` - Raw provider API responses for debugging
+
+## Environment Variables
+
+Optional API keys for fetching provider-specific data:
+- `NOVITA_API_KEY`
+- `SAMBANOVA_API_KEY`
+- `GROQ_API_KEY`
+- `FEATHERLESS_API_KEY`
+- `TOGETHER_API_KEY`
+- `COHERE_API_KEY`
+- `FIREWORKS_API_KEY`
+- `HF_TOKEN` - Required for performance testing
+
+This project was created using `bun init` in bun v1.2.4. [Bun](https://bun.sh) is a fast all-in-one JavaScript runtime.

bun.lock

@@ -0,0 +1,29 @@
+{
+  "lockfileVersion": 1,
+  "workspaces": {
+    "": {
+      "name": "llm-pricing",
+      "devDependencies": {
+        "@types/bun": "latest",
+      },
+      "peerDependencies": {
+        "typescript": "^5",
+      },
+    },
+  },
+  "packages": {
+    "@types/bun": ["@types/[email protected]", "", { "dependencies": { "bun-types": "1.2.19" } }, "sha512-d9ZCmrH3CJ2uYKXQIUuZ/pUnTqIvLDS0SK7pFmbx8ma+ziH/FRMoAq5bYpRG7y+w1gl+HgyNZbtqgMq4W4e2Lg=="],
+
+    "@types/node": ["@types/[email protected]", "", { "dependencies": { "undici-types": "~7.8.0" } }, "sha512-ut5FthK5moxFKH2T1CUOC6ctR67rQRvvHdFLCD2Ql6KXmMuCrjsSsRI9UsLCm9M18BMwClv4pn327UvB7eeO1w=="],
+
+    "@types/react": ["@types/[email protected]", "", { "dependencies": { "csstype": "^3.0.2" } }, "sha512-WmdoynAX8Stew/36uTSVMcLJJ1KRh6L3IZRx1PZ7qJtBqT3dYTgyDTx8H1qoRghErydW7xw9mSJ3wS//tCRpFA=="],
+
+    "bun-types": ["[email protected]", "", { "dependencies": { "@types/node": "*" }, "peerDependencies": { "@types/react": "^19" } }, "sha512-uAOTaZSPuYsWIXRpj7o56Let0g/wjihKCkeRqUBhlLVM/Bt+Fj9xTo+LhC1OV1XDaGkz4hNC80et5xgy+9KTHQ=="],
+
+    "csstype": ["[email protected]", "", {}, "sha512-M1uQkMl8rQK/szD0LNhtqxIPLpimGm8sOBwU7lLnCpSbTyY3yeU1Vc7l4KT5zT4s/yOxHH5O7tIuuLOCnLADRw=="],
+
+    "typescript": ["[email protected]", "", { "bin": { "tsc": "bin/tsc", "tsserver": "bin/tsserver" } }, "sha512-p1diW6TqL9L07nNxvRMM7hMMw4c5XOo/1ibL4aAIGmSAt9slTE1Xgw5KWuof2uTOvCg9BY7ZRi+GaF+7sfgPeQ=="],
+
+    "undici-types": ["[email protected]", "", {}, "sha512-9UJ2xGDvQ43tYyVMpuHlsgApydB8ZKfVYTsLDhXkFL/6gfkp+U8xTGdh8pMJv1SpZna0zxG1DwsKZsreLbXBxw=="],
+  }
+}

get-metrics.ts

@@ -0,0 +1,611 @@
+// get-metrics-new.ts - Updated version using direct provider APIs
+import * as fs from "node:fs";
+import { parseArgs } from "util";
+import { ProviderAggregator } from "./providers";
+import type { ProviderEntry } from "./providers";
+import { extractHFRouterData } from "./providers/huggingface-router";
+
+/* -------------------------------------------------------------------------- */
+/*  CONSTANTS                                                                 */
+/* -------------------------------------------------------------------------- */
+
+const HUGGINGFACE_API = "https://router.huggingface.co/v1/models";
+const HUGGINGFACE_ROUTER_API =
+  "https://router.huggingface.co/v1/chat/completions";
+
+/* -------------------------------------------------------------------------- */
+/*  TYPE DEFINITIONS                                                          */
+/* -------------------------------------------------------------------------- */
+
+interface HFModel {
+  id: string;
+  [key: string]: any;
+  providers?: ProviderEntry[];
+}
+
+interface Statistics {
+  total_models: number;
+  models_enriched: number;
+  providers_enriched: number;
+  new_capabilities_added: number;
+  providers_fetched: Record<string, number>;
+}
+
+interface PerformanceTestResult {
+  total_tested: number;
+  successful: number;
+  errors: number;
+  status_distribution: Record<string, number>;
+}
+
+/* -------------------------------------------------------------------------- */
+/*  FETCH HELPERS                                                             */
+/* -------------------------------------------------------------------------- */
+
+async function fetchHuggingfaceModels(): Promise<HFModel[]> {
+  const resp = await fetch(HUGGINGFACE_API).then(
+    (r) => r.json() as Promise<{ data: HFModel[] }>
+  );
+  return resp.data;
+}
+
+/* -------------------------------------------------------------------------- */
+/*  PROVIDER ENRICHMENT                                                       */
+/* -------------------------------------------------------------------------- */
+
+function normalizeModelId(modelId: string): string {
+  // Convert HF model ID to a normalized form for matching
+  // Remove organization prefix for common patterns
+  const patterns = [
+    /^meta-llama\/Meta-Llama-(.+)$/,
+    /^meta-llama\/Llama-(.+)$/,
+    /^mistralai\/(.+)$/,
+    /^google\/(.+)$/,
+    /^anthropic\/(.+)$/,
+  ];
+
+  for (const pattern of patterns) {
+    const match = modelId.match(pattern);
+    if (match) {
+      return match[1].toLowerCase();
+    }
+  }
+
+  // For other models, just use the part after the last slash
+  const parts = modelId.split("/");
+  return parts[parts.length - 1].toLowerCase();
+}
+
+function matchProviderModel(
+  hfModelId: string,
+  providerEntries: Map<string, ProviderEntry[]>
+): Map<string, ProviderEntry[]> {
+  const normalizedHfId = normalizeModelId(hfModelId);
+  const matches = new Map<string, ProviderEntry[]>();
+
+  for (const [provider, entries] of providerEntries) {
+    const matchingEntries = entries.filter((entry) => {
+      // This would need to be enhanced with provider-specific matching logic
+      // For now, we'll use simple substring matching
+      const entryId = (entry as any).id || (entry as any).model_id || "";
+      const normalizedEntryId = normalizeModelId(entryId);
+
+      return (
+        normalizedEntryId.includes(normalizedHfId) ||
+        normalizedHfId.includes(normalizedEntryId)
+      );
+    });
+
+    if (matchingEntries.length > 0) {
+      matches.set(provider, matchingEntries);
+    }
+  }
+
+  return matches;
+}
+
+async function enrichHuggingfaceModels(
+  hfModels: HFModel[],
+  aggregator: ProviderAggregator
+): Promise<{
+  enriched: HFModel[];
+  stats: Statistics;
+  matchedProviderData: any[];
+}> {
+  console.log("\nFetching data from all providers...");
+  const providerData = await aggregator.fetchAllProviders();
+
+  const stats: Statistics = {
+    total_models: hfModels.length,
+    models_enriched: 0,
+    providers_enriched: 0,
+    new_capabilities_added: 0,
+    providers_fetched: {},
+  };
+
+  // Count models per provider
+  for (const [provider, entries] of providerData) {
+    stats.providers_fetched[provider] = entries.length;
+  }
+
+  const enrichedModels: HFModel[] = [];
+  const matchedProviderData: any[] = [];
+  const matchedProviderKeys = new Set<string>(); // Track unique model-provider combinations
+
+  console.log(
+    `\nProcessing ${hfModels.length} models from HuggingFace Router API...`
+  );
+
+  for (const hfModel of hfModels) {
+    const enrichedModel = structuredClone(hfModel);
+
+    // Extract HF router data first (this is already in the model)
+    const hfRouterData = extractHFRouterData(enrichedModel);
+
+    // Find matches from provider APIs
+    const matches = matchProviderModel(hfModel.id, providerData);
+
+    // Ensure providers array exists
+    if (!enrichedModel.providers) {
+      enrichedModel.providers = [];
+    }
+
+    let modelEnriched = false;
+
+    // Process HF router data first (prioritize it)
+    for (const [providerName, hfProviderData] of hfRouterData) {
+      const normalizedProvider = normalizeProviderName(providerName);
+
+      // Check if provider already exists in the model
+      let existingProvider = enrichedModel.providers.find(
+        (p) => normalizeProviderName(p.provider) === normalizedProvider
+      );
+
+      if (existingProvider) {
+        // HF router data is already there, just count it
+        if (hfProviderData.pricing) {
+          stats.providers_enriched++;
+          modelEnriched = true;
+        }
+        // Track this provider data as matched (avoid duplicates)
+        const matchKey = `${hfModel.id}:${providerName}`;
+        if (!matchedProviderKeys.has(matchKey)) {
+          matchedProviderKeys.add(matchKey);
+          matchedProviderData.push({
+            ...hfProviderData,
+            provider: providerName,
+            id: hfModel.id,
+          });
+        }
+      }
+    }
+
+    // Then enrich with provider API data where missing
+    if (matches.size > 0) {
+      for (const [provider, providerEntries] of matches) {
+        for (const providerEntry of providerEntries) {
+          // Find existing provider entry
+          let existingProvider = enrichedModel.providers.find(
+            (p) => normalizeProviderName(p.provider) === provider.toLowerCase()
+          );
+
+          if (!existingProvider) {
+            // No HF router data for this provider
+            // Skip - we only want providers that are listed in HF Router
+            continue;
+          } else {
+            // Merge data, but prioritize HF router data
+            const hadPricing = !!existingProvider.pricing;
+            const hadTools = existingProvider.supports_tools !== undefined;
+            const hadStructured =
+              existingProvider.supports_structured_output !== undefined;
+            const hadContext = !!existingProvider.context_length;
+
+            // Only add provider API data for missing fields
+            const mergedData: any = {};
+
+            // Add provider API data only if HF router doesn't have it
+            if (!hadPricing && providerEntry.pricing) {
+              mergedData.pricing = providerEntry.pricing;
+              stats.providers_enriched++;
+              modelEnriched = true;
+            }
+
+            if (!hadContext && providerEntry.context_length) {
+              mergedData.context_length = providerEntry.context_length;
+            }
+
+            if (!hadTools && providerEntry.supports_tools !== undefined) {
+              mergedData.supports_tools = providerEntry.supports_tools;
+            }
+
+            if (
+              !hadStructured &&
+              providerEntry.supports_structured_output !== undefined
+            ) {
+              mergedData.supports_structured_output =
+                providerEntry.supports_structured_output;
+            }
+
+            // Add other capabilities from provider API
+            for (const key of Object.keys(providerEntry)) {
+              if (
+                key.startsWith("supports_") &&
+                !["supports_tools", "supports_structured_output"].includes(
+                  key
+                ) &&
+                !(key in existingProvider)
+              ) {
+                mergedData[key] = (providerEntry as any)[key];
+                stats.new_capabilities_added++;
+              }
+            }
+
+            // Apply merged data
+            Object.assign(existingProvider, mergedData);
+
+            // Track the enriched data (avoid duplicates)
+            const matchKey = `${hfModel.id}:${provider}`;
+            if (!matchedProviderKeys.has(matchKey)) {
+              matchedProviderKeys.add(matchKey);
+              matchedProviderData.push({
+                ...existingProvider,
+                provider,
+                id: hfModel.id,
+              });
+            }
+          }
+        }
+      }
+    }
+
+    if (modelEnriched) {
+      stats.models_enriched++;
+    }
+
+    enrichedModels.push(enrichedModel);
+  }
+
+  // Log models from provider APIs that weren't matched
+  let unmatchedCount = 0;
+  for (const [provider, entries] of providerData) {
+    for (const entry of entries) {
+      const modelId = (entry as any).model_id || (entry as any).id || "";
+      if (modelId) {
+        const matchKey = `${modelId}:${provider}`;
+        if (!matchedProviderKeys.has(matchKey)) {
+          unmatchedCount++;
+        }
+      }
+    }
+  }
+
+  if (unmatchedCount > 0) {
+    console.log(
+      `\nNote: ${unmatchedCount} models from provider APIs were not included (not in HF Router).`
+    );
+  }
+
+  return { enriched: enrichedModels, stats, matchedProviderData };
+}
+
+// Helper function to normalize provider names for comparison
+function normalizeProviderName(providerName: string): string {
+  const providerMap: Record<string, string> = {
+    "featherless-ai": "featherless",
+    "fireworks-ai": "fireworks",
+    "hf-inference": "huggingface",
+  };
+
+  return (providerMap[providerName] || providerName).toLowerCase();
+}
+
+/* -------------------------------------------------------------------------- */
+/*  PERFORMANCE TESTING                                                       */
+/* -------------------------------------------------------------------------- */
+
+async function testModelProvider(
+  modelId: string,
+  providerName: string,
+  hfToken: string
+): Promise<Partial<ProviderEntry>> {
+  const nonce = crypto.randomUUID().slice(0, 8);
+  const prompt = `What is the capital of France?\n<!-- nonce:${nonce} -->`;
+
+  const payload = {
+    model: `${modelId}:${providerName}`,
+    messages: [{ role: "user", content: prompt }],
+    stream: false,
+    temperature: 0.7,
+  };
+
+  const headers = {
+    Authorization: `Bearer ${hfToken}`,
+    "Content-Type": "application/json",
+  };
+
+  const start = performance.now();
+  try {
+    const controller = new AbortController();
+    const timeoutId = setTimeout(() => controller.abort(), 30_000);
+
+    const resp = await fetch(HUGGINGFACE_ROUTER_API, {
+      method: "POST",
+      headers,
+      body: JSON.stringify(payload),
+      signal: controller.signal,
+    });
+
+    clearTimeout(timeoutId);
+    const latency = (performance.now() - start) / 1000;
+
+    if (resp.ok) {
+      const data = await resp.json();
+      const usage = data.usage ?? {};
+      const totalTokens =
+        usage.total_tokens ??
+        (usage.prompt_tokens ?? 0) + (usage.completion_tokens ?? 0);
+      const tps = totalTokens ? totalTokens / latency : 0;
+
+      return {
+        latency_s: Number(latency.toFixed(2)),
+        throughput_tps: Number(tps.toFixed(2)),
+        status: "live",
+      };
+    }
+
+    const data = await resp.json().catch(() => ({}));
+    const msg =
+      data?.error?.message ?? `HTTP ${resp.status} ${resp.statusText}`;
+    return { performance_error: msg, status: "offline" };
+  } catch (err: any) {
+    const msg = err.name === "AbortError" ? "Request timeout" : err.message;
+    return { performance_error: msg, status: "offline" };
+  }
+}
+
+async function testProvidersBatch(
+  triplets: [string, string, ProviderEntry][],
+  hfToken: string
+): Promise<void> {
+  await Promise.all(
+    triplets.map(async ([modelId, providerName, prov]) => {
+      const res = await testModelProvider(modelId, providerName, hfToken);
+      Object.assign(prov, res, {
+        performance_tested_at: new Date().toISOString(),
+      });
+    })
+  );
+}
+
+async function testAllProviders(
+  models: HFModel[],
+  hfToken: string,
+  limit: number | undefined,
+  batchSize: number,
+  filter: string[] | undefined
+): Promise<PerformanceTestResult> {
+  const subset = typeof limit === "number" ? models.slice(0, limit) : models;
+
+  const allPairs: [string, string, ProviderEntry][] = [];
+  for (const m of subset) {
+    for (const p of m.providers ?? []) {
+      if (filter && !filter.includes(p.provider)) continue;
+      allPairs.push([m.id, p.provider, p]);
+    }
+  }
+
+  console.log(
+    `\nTesting performance for ${allPairs.length} model-provider combinations...`
+  );
+
+  let tested = 0;
+  let errors = 0;
+  const statusDist: Record<string, number> = {
+    live: 0,
+    offline: 0,
+    not_tested: 0,
+  };
+
+  for (let i = 0; i < allPairs.length; i += batchSize) {
+    const batch = allPairs.slice(i, i + batchSize);
+    console.log(
+      `Testing batch ${i / batchSize + 1}/${Math.ceil(
+        allPairs.length / batchSize
+      )}...`
+    );
+    await testProvidersBatch(batch, hfToken);
+
+    batch.forEach(([_, __, prov]) => {
+      tested += 1;
+      if (prov.performance_error) errors += 1;
+      switch (prov.status) {
+        case "live":
+          statusDist.live += 1;
+          break;
+        case "offline":
+          statusDist.offline += 1;
+          break;
+        default:
+          statusDist.not_tested += 1;
+      }
+    });
+
+    if (i + batchSize < allPairs.length) {
+      await new Promise((resolve) => setTimeout(resolve, 1000));
+    }
+  }
+
+  return {
+    total_tested: tested,
+    successful: tested - errors,
+    errors,
+    status_distribution: statusDist,
+  };
+}
+
+/* -------------------------------------------------------------------------- */
+/*  PRINT HELPERS                                                             */
+/* -------------------------------------------------------------------------- */
+
+function printStatistics(s: Statistics): void {
+  console.log("\n" + "=".repeat(60));
+  console.log("ENRICHMENT STATISTICS");
+  console.log("=".repeat(60));
+  console.log(`Total models processed: ${s.total_models}`);
+  console.log(`Models enriched with pricing: ${s.models_enriched}`);
+  console.log(`Provider entries enriched: ${s.providers_enriched}`);
+  console.log(`New capability fields added: ${s.new_capabilities_added}`);
+
+  console.log("\nProvider data fetched:");
+  Object.entries(s.providers_fetched)
+    .sort(([a], [b]) => a.localeCompare(b))
+    .forEach(([provider, count]) => {
+      console.log(`  ${provider}: ${count} models`);
+    });
+}
+
+/* -------------------------------------------------------------------------- */
+/*  CLI PARSER                                                                */
+/* -------------------------------------------------------------------------- */
+
+const { values: opts } = parseArgs({
+  args: Bun.argv.slice(2),
+  options: {
+    "test-performance": { type: "boolean" },
+    "test-limit": { type: "string" },
+    "test-providers": { type: "string", multiple: true },
+    "batch-size": { type: "string" },
+    providers: { type: "string", multiple: true },
+    "skip-providers": { type: "string", multiple: true },
+  },
+  strict: false,
+});
+
+const testLimit =
+  opts["test-limit"] && typeof opts["test-limit"] === "string"
+    ? parseInt(opts["test-limit"], 10)
+    : undefined;
+const batchSize =
+  opts["batch-size"] && typeof opts["batch-size"] === "string"
+    ? parseInt(opts["batch-size"], 10)
+    : 20;
+
+/* -------------------------------------------------------------------------- */
+/*  MAIN                                                                      */
+/* -------------------------------------------------------------------------- */
+
+(async () => {
+  console.log("Fetching HuggingFace models...");
+  const hfModels = await fetchHuggingfaceModels();
+  console.log(`Found ${hfModels.length} HuggingFace models.`);
+
+  // Configure provider aggregator
+  const apiKeys: Record<string, string> = {};
+  
+  // Only add API keys that are defined
+  if (process.env.NOVITA_API_KEY) apiKeys.novita = process.env.NOVITA_API_KEY;
+  if (process.env.SAMBANOVA_API_KEY) apiKeys.sambanova = process.env.SAMBANOVA_API_KEY;
+  if (process.env.GROQ_API_KEY) apiKeys.groq = process.env.GROQ_API_KEY;
+  if (process.env.FEATHERLESS_API_KEY) apiKeys.featherless = process.env.FEATHERLESS_API_KEY;
+  if (process.env.TOGETHER_API_KEY) apiKeys.together = process.env.TOGETHER_API_KEY;
+  if (process.env.COHERE_API_KEY) apiKeys.cohere = process.env.COHERE_API_KEY;
+  if (process.env.FIREWORKS_API_KEY) apiKeys.fireworks = process.env.FIREWORKS_API_KEY;
+  if (process.env.NEBIUS_API_KEY) apiKeys.nebius = process.env.NEBIUS_API_KEY;
+  if (process.env.HYPERBOLIC_API_KEY) apiKeys.hyperbolic = process.env.HYPERBOLIC_API_KEY;
+  if (process.env.CEREBRAS_API_KEY) apiKeys.cerebras = process.env.CEREBRAS_API_KEY;
+  if (process.env.NSCALE_API_KEY) apiKeys.nscale = process.env.NSCALE_API_KEY;
+  
+  const config = {
+    providers: opts["providers"] as string[] | undefined,
+    apiKeys,
+  };
+
+  // Remove skip-providers if specified
+  if (opts["skip-providers"]) {
+    const skipProviders = opts["skip-providers"] as string[];
+    if (!config.providers) {
+      config.providers = [
+        "novita",
+        "sambanova",
+        "groq",
+        "featherless",
+        "together",
+        "cohere",
+        "fireworks",
+        "nebius",
+        "hyperbolic",
+        "cerebras",
+        "nscale",
+      ].filter((p) => !skipProviders.includes(p));
+    }
+  }
+
+  const aggregator = new ProviderAggregator(config);
+
+  console.log("\nEnriching HuggingFace models with provider data...");
+  const { enriched, stats, matchedProviderData } =
+    await enrichHuggingfaceModels(hfModels, aggregator);
+
+  // Optional performance tests
+  if (opts["test-performance"]) {
+    const hfToken = process.env.HF_TOKEN;
+    if (!hfToken) {
+      console.error(
+        "ERROR: HF_TOKEN environment variable not set. Skipping performance tests."
+      );
+    } else {
+      console.log("\n" + "=".repeat(60));
+      console.log("PERFORMANCE TESTING");
+      console.log("=".repeat(60));
+      const perfStats = await testAllProviders(
+        enriched,
+        hfToken,
+        testLimit,
+        batchSize,
+        opts["test-providers"] as string[] | undefined
+      );
+      console.log("\nPerformance testing complete:");
+      console.log(`  Total tested: ${perfStats.total_tested}`);
+      console.log(`  Successful: ${perfStats.successful}`);
+      console.log(`  Errors: ${perfStats.errors}`);
+      console.log("\nProvider status distribution:");
+      Object.entries(perfStats.status_distribution)
+        .sort()
+        .forEach(([k, v]) => console.log(`  ${k}: ${v}`));
+    }
+  }
+
+  // Save enriched data
+  const outFile = "enriched_models_enhanced.json";
+  fs.writeFileSync(
+    outFile,
+    JSON.stringify(
+      {
+        data: enriched,
+        generated_at: new Date().toISOString(),
+        metadata: {
+          total_models: enriched.length,
+          models_enriched: stats.models_enriched,
+          providers_enriched: stats.providers_enriched,
+          performance_tested: !!opts["test-performance"],
+          providers_fetched: stats.providers_fetched,
+        },
+      },
+      null,
+      2
+    )
+  );
+  console.log(`\nEnriched data saved → ${outFile}`);
+
+  // Save only matched provider data (models that exist in HF Router)
+  fs.writeFileSync(
+    "provider_models_raw.json",
+    JSON.stringify({ data: matchedProviderData }, null, 2)
+  );
+  console.log(
+    `Matched provider models saved → provider_models_raw.json (${matchedProviderData.length} entries)`
+  );
+
+  printStatistics(stats);
+})();

index.html

@@ -0,0 +1,332 @@
+<!DOCTYPE html>
+<html>
+<head>
+    <title>HuggingFace Models - Enriched</title>
+    <meta charset="UTF-8">
+    <style>
+        body {
+            font-family: monospace;
+            margin: 20px;
+        }
+        input {
+            font-family: monospace;
+            border: 1px solid #000;
+            padding: 4px 8px;
+            width: 300px;
+        }
+        table {
+            border-collapse: collapse;
+            width: 100%;
+        }
+        thead {
+            position: sticky;
+            top: 0;
+            z-index: 10;
+        }
+        th, td {
+            border: 1px solid #000;
+            padding: 4px 8px;
+            text-align: left;
+        }
+        tr.model-group-start td {
+            border-top: 2px solid #000;
+        }
+        th {
+            background: #f0f0f0;
+            font-weight: bold;
+            cursor: pointer;
+            user-select: none;
+            position: relative;
+        }
+        th:hover {
+            background: #e0e0e0;
+        }
+        th::after {
+            content: ' ↕';
+            color: #999;
+            font-size: 0.8em;
+        }
+        th.sort-asc::after {
+            content: ' ↑';
+            color: #333;
+        }
+        th.sort-desc::after {
+            content: ' ↓';
+            color: #333;
+        }
+        tr:hover {
+            background: #f9f9f9;
+        }
+        .hidden {
+            display: none;
+        }
+        .highlighted {
+            background: #fffacd !important;
+        }
+        .best-value {
+            color: #008000;
+            font-weight: bold;
+        }
+        .header-container {
+            display: flex;
+            justify-content: space-between;
+            align-items: center;
+            gap: 20px;
+            margin-bottom: 10px;
+        }
+        .generation-date {
+            color: #666;
+            font-size: 0.9em;
+        }
+    </style>
+</head>
+<body>
+    <div class="header-container">
+        <input type="search" id="filterInput" placeholder="Filter by model or provider...">
+        <span class="generation-date" id="generationDate"></span>
+    </div>
+    <table id="modelsTable">
+        <thead>
+            <tr>
+                <th>Model</th>
+                <th>Provider</th>
+                <th>Status</th>
+                <th>Uptime %</th>
+                <th>Input $/1M</th>
+                <th>Output $/1M</th>
+                <th>Context</th>
+                <th>Quant</th>
+                <th>Latency (s)</th>
+                <th>Throughput (t/s)</th>
+                <th>Tools</th>
+                <th>Structured</th>
+            </tr>
+        </thead>
+        <tbody id="tableBody">
+            <tr><td colspan="12">Loading...</td></tr>
+        </tbody>
+    </table>
+
+    <script>
+        // Get query parameters
+        const urlParams = new URLSearchParams(window.location.search);
+        const highlightModelId = urlParams.get('model');
+        
+        fetch('enriched_models_enhanced.json')
+            .then(response => response.json())
+            .then(data => {
+                // Display generation date
+                if (data.generated_at) {
+                    const date = new Date(data.generated_at);
+                    const dateStr = date.toLocaleString('en-US', {
+                        year: 'numeric',
+                        month: 'short',
+                        day: 'numeric',
+                        hour: '2-digit',
+                        minute: '2-digit',
+                        timeZoneName: 'short'
+                    });
+                    document.getElementById('generationDate').textContent = `Last update: ${dateStr}`;
+                }
+                
+                const tbody = document.getElementById('tableBody');
+                tbody.innerHTML = '';
+                let firstHighlightedRow = null;
+                
+                // Handle both old format (direct array) and new format (with metadata)
+                const models = Array.isArray(data) ? data : data.data;
+                models.forEach((model, modelIndex) => {
+                    if (model.providers) {
+                        model.providers.forEach((provider, providerIndex) => {
+                            const row = document.createElement('tr');
+                            
+                            // Add class for first provider of each model to create visual separation
+                            if (providerIndex === 0 && modelIndex > 0) {
+                                row.classList.add('model-group-start');
+                            }
+                            
+                            // Highlight if model matches query parameter
+                            if (highlightModelId && model.id === highlightModelId) {
+                                row.classList.add('highlighted');
+                                if (!firstHighlightedRow) {
+                                    firstHighlightedRow = row;
+                                }
+                            }
+                    
+                    row.innerHTML = `
+                        <td>${model.id}</td>
+                        <td>${provider.provider}</td>
+                        <td>${provider.endpoint_status_name || provider.status || '-'}</td>
+                        <td>${provider.uptime_30d !== undefined ? provider.uptime_30d : '-'}</td>
+                        <td>${provider.pricing?.input !== undefined ? provider.pricing.input : '-'}</td>
+                        <td>${provider.pricing?.output !== undefined ? provider.pricing.output : '-'}</td>
+                        <td>${provider.context_length || '-'}</td>
+                        <td>${provider.quantization || '-'}</td>
+                        <td>${provider.latency_s !== undefined ? provider.latency_s : '-'}</td>
+                        <td>${provider.throughput_tps !== undefined ? provider.throughput_tps : '-'}</td>
+                        <td>${provider.supports_tools ? 'Yes' : 'No'}</td>
+                        <td>${provider.supports_structured_output ? 'Yes' : 'No'}</td>
+                    `;
+                    
+                    tbody.appendChild(row);
+                        });
+                    }
+                });
+                
+                // Store original data for sorting
+                window.tableData = models;
+                
+                // Function to find and mark best values
+                function markBestValues() {
+                    const rows = Array.from(tbody.getElementsByTagName('tr'));
+                    const highlightedRows = rows.filter(row => row.classList.contains('highlighted'));
+                    
+                    if (highlightedRows.length === 0) return;
+                    
+                    // Define which columns need min vs max for best value
+                    const columnConfig = {
+                        4: 'min', // Input $/1M - lower is better
+                        5: 'min', // Output $/1M - lower is better
+                        6: 'max', // Context - higher is better
+                        8: 'min', // Latency - lower is better
+                        9: 'max', // Throughput - higher is better
+                        3: 'max'  // Uptime % - higher is better
+                    };
+                    
+                    // For each configured column, find the best value among highlighted rows
+                    Object.entries(columnConfig).forEach(([colIndex, type]) => {
+                        const values = highlightedRows
+                            .map(row => {
+                                const cellText = row.cells[colIndex].textContent.trim();
+                                const value = cellText === '-' ? null : parseFloat(cellText);
+                                return { row, value, cell: row.cells[colIndex] };
+                            })
+                            .filter(item => item.value !== null && !isNaN(item.value));
+                        
+                        if (values.length === 0) return;
+                        
+                        // Find best value
+                        let bestValue;
+                        if (type === 'min') {
+                            bestValue = Math.min(...values.map(v => v.value));
+                        } else {
+                            bestValue = Math.max(...values.map(v => v.value));
+                        }
+                        
+                        // Mark cells with best value
+                        values.forEach(item => {
+                            if (item.value === bestValue) {
+                                item.cell.classList.add('best-value');
+                            }
+                        });
+                    });
+                }
+                
+                // Call markBestValues if model is highlighted
+                if (highlightModelId) {
+                    markBestValues();
+                }
+                
+                // Scroll to highlighted model if present
+                if (firstHighlightedRow) {
+                    setTimeout(() => {
+                        firstHighlightedRow.scrollIntoView({ behavior: 'smooth', block: 'center' });
+                    }, 100);
+                }
+                
+                // Filter functionality
+                document.getElementById('filterInput').addEventListener('input', function(e) {
+                    const filter = e.target.value.toLowerCase();
+                    const rows = tbody.getElementsByTagName('tr');
+                    
+                    for (let row of rows) {
+                        const modelText = row.cells[0].textContent.toLowerCase();
+                        const providerText = row.cells[1].textContent.toLowerCase();
+                        
+                        if (modelText.includes(filter) || providerText.includes(filter)) {
+                            row.classList.remove('hidden');
+                        } else {
+                            row.classList.add('hidden');
+                        }
+                    }
+                });
+                
+                // Sorting functionality
+                let sortColumn = -1;
+                let sortDirection = 'asc';
+                
+                const headers = document.querySelectorAll('th');
+                headers.forEach((header, index) => {
+                    header.addEventListener('click', () => {
+                        // Remove sort classes from all headers
+                        headers.forEach(h => {
+                            h.classList.remove('sort-asc', 'sort-desc');
+                        });
+                        
+                        // Determine sort direction
+                        if (sortColumn === index) {
+                            sortDirection = sortDirection === 'asc' ? 'desc' : 'asc';
+                        } else {
+                            sortColumn = index;
+                            sortDirection = 'asc';
+                        }
+                        
+                        // Add sort class to current header
+                        header.classList.add(sortDirection === 'asc' ? 'sort-asc' : 'sort-desc');
+                        
+                        // Sort the table
+                        sortTable(index, sortDirection);
+                    });
+                });
+                
+                function sortTable(columnIndex, direction) {
+                    const rows = Array.from(tbody.getElementsByTagName('tr'));
+                    
+                    rows.sort((a, b) => {
+                        const aText = a.cells[columnIndex].textContent.trim();
+                        const bText = b.cells[columnIndex].textContent.trim();
+                        
+                        // Handle special cases
+                        if (aText === '-' && bText !== '-') return direction === 'asc' ? 1 : -1;
+                        if (aText !== '-' && bText === '-') return direction === 'asc' ? -1 : 1;
+                        if (aText === '-' && bText === '-') return 0;
+                        
+                        // Try to parse as number
+                        const aNum = parseFloat(aText);
+                        const bNum = parseFloat(bText);
+                        
+                        let comparison = 0;
+                        if (!isNaN(aNum) && !isNaN(bNum)) {
+                            comparison = aNum - bNum;
+                        } else {
+                            // Handle Yes/No specially
+                            if (aText === 'Yes' || aText === 'No') {
+                                comparison = aText === bText ? 0 : (aText === 'Yes' ? -1 : 1);
+                            } else {
+                                comparison = aText.localeCompare(bText);
+                            }
+                        }
+                        
+                        return direction === 'asc' ? comparison : -comparison;
+                    });
+                    
+                    // Clear tbody and re-append sorted rows
+                    tbody.innerHTML = '';
+                    rows.forEach((row, index) => {
+                        // Re-apply model-group-start class based on model changes
+                        if (index > 0 && rows[index].cells[0].textContent !== rows[index-1].cells[0].textContent) {
+                            row.classList.add('model-group-start');
+                        } else if (index > 0) {
+                            row.classList.remove('model-group-start');
+                        }
+                        tbody.appendChild(row);
+                    });
+                }
+            })
+            .catch(error => {
+                console.error('Error loading data:', error);
+                document.getElementById('tableBody').innerHTML = '<tr><td colspan="12">Error loading data</td></tr>';
+            });
+    </script>
+</body>
+</html>

package.json

@@ -0,0 +1,10 @@
+{
+  "name": "llm-pricing",
+  "private": true,
+  "devDependencies": {
+    "@types/bun": "latest"
+  },
+  "peerDependencies": {
+    "typescript": "^5"
+  }
+}

provider_models_raw.json

@@ -0,0 +1,2417 @@
+{
+  "data": [
+    {
+      "provider": "novita",
+      "status": "live",
+      "id": "zai-org/GLM-4.5"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.95,
+        "output": 5
+      },
+      "context_length": 262144,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 2,
+        "output": 2
+      },
+      "context_length": 262144,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "id": "moonshotai/Kimi-K2-Instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "moonshotai/Kimi-K2-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "moonshotai/Kimi-K2-Instruct"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "id": "moonshotai/Kimi-K2-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "moonshotai/Kimi-K2-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "moonshotai/Kimi-K2-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.3,
+        "output": 3
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-235B-A22B-Thinking-2507"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-235B-A22B-Thinking-2507"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "CohereLabs/command-a-vision-07-2025"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.15,
+        "output": 0.8
+      },
+      "context_length": 262144,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.6
+      },
+      "context_length": 262144,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.02,
+        "output": 0.05
+      },
+      "context_length": 16384,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.03,
+        "output": 0.09
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "offline",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.2
+      },
+      "context_length": 16384,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "meta-llama/Llama-3.1-8B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "zai-org/GLM-4.5-Air-FP8"
+    },
+    {
+      "provider": "hf-inference",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "HuggingFaceTB/SmolLM3-3B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.7,
+        "output": 2.5
+      },
+      "context_length": 64000,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 2,
+        "output": 6
+      },
+      "context_length": 163840,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 3,
+        "output": 7
+      },
+      "context_length": 163840,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "deepseek-ai/DeepSeek-R1"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.45
+      },
+      "context_length": 40960,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-30B-A3B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.3,
+        "output": 0.9
+      },
+      "context_length": 40960,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-30B-A3B"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-30B-A3B"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "Qwen/Qwen3-30B-A3B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen3-8B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-8B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen3-8B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.6
+      },
+      "context_length": 110000,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "google/gemma-3-27b-it"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "google/gemma-3-27b-it"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "google/gemma-3-27b-it"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.035,
+        "output": 0.138
+      },
+      "context_length": 65536,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "zai-org/GLM-4.1V-9B-Thinking"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 2,
+        "output": 2
+      },
+      "context_length": 262144,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-0528"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.8,
+        "output": 2.4
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-0528"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-0528"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-0528"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-0528"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-0528"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "context_length": 163840,
+      "supports_tools": false,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "deepseek-ai/DeepSeek-R1-0528"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2.5-VL-7B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.04,
+        "output": 0.04
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Meta-Llama-3-8B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "meta-llama/Meta-Llama-3-8B-Instruct"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.05,
+        "output": 0.08
+      },
+      "context_length": 8192,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Meta-Llama-3-8B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "meta-llama/Meta-Llama-3-8B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.03,
+        "output": 0.05
+      },
+      "context_length": 32768,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.2-3B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.060000000000000005,
+        "output": 0.060000000000000005
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.2-3B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.2-3B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.45
+      },
+      "context_length": 40960,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-32B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.3
+      },
+      "context_length": 40960,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-32B"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "pricing": {
+        "input": 0.4,
+        "output": 0.8
+      },
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-32B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen3-32B"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.29,
+        "output": 0.59
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-32B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-32B"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "pricing": {
+        "input": 0.4,
+        "output": 0.8
+      },
+      "context_length": 32768,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-32B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen3-32B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.06,
+        "output": 0.09
+      },
+      "context_length": 128000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1-0528-Qwen3-8B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-0528-Qwen3-8B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "deepseek-ai/DeepSeek-R1-0528-Qwen3-8B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.28,
+        "output": 1.14
+      },
+      "context_length": 163840,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-V3-0324"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 2,
+        "output": 6
+      },
+      "context_length": 163840,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-V3-0324"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-V3-0324"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 1.25,
+        "output": 1.25
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-V3-0324"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-V3-0324"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "pricing": {
+        "input": 3,
+        "output": 4.5
+      },
+      "context_length": 32768,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-V3-0324"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "deepseek-ai/DeepSeek-V3-0324"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.13,
+        "output": 0.39
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.25,
+        "output": 0.75
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "pricing": {
+        "input": 0.85,
+        "output": 1.2
+      },
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.88,
+        "output": 0.88
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.59,
+        "output": 0.79
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "pricing": {
+        "input": 0.6,
+        "output": 1.2
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "meta-llama/Llama-3.3-70B-Instruct"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.02,
+        "output": 0.06
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "google/gemma-2-2b-it"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.24,
+        "output": 0.24
+      },
+      "context_length": 32000,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "zai-org/GLM-4-32B-0414"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "zai-org/GLM-4-32B-0414"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "zai-org/GLM-4-32B-0414"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.08,
+        "output": 0.24
+      },
+      "context_length": 40960,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-4B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.5
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "pricing": {
+        "input": 0.65,
+        "output": 0.85
+      },
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.18000000000000002,
+        "output": 0.5900000000000001
+      },
+      "context_length": 1048576,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.11,
+        "output": 0.34
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "context_length": 131000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.2-1B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen2.5-7B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.3,
+        "output": 0.3
+      },
+      "context_length": 32768,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2.5-7B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen2.5-7B-Instruct"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.08,
+        "output": 0.24
+      },
+      "context_length": 40960,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-14B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen3-14B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen3-14B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.07,
+        "output": 0.28
+      },
+      "context_length": 120000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "baidu/ERNIE-4.5-21B-A3B-PT"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.13,
+        "output": 0.4
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.1-70B-Instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.1-70B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "offline",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.1-70B-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "meta-llama/Llama-3.1-70B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.14,
+        "output": 0.56
+      },
+      "context_length": 30000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "baidu/ERNIE-4.5-VL-28B-A3B-PT"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.3
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2.5-Coder-32B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen2.5-Coder-32B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.8,
+        "output": 0.8
+      },
+      "context_length": 16384,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-Coder-32B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2.5-Coder-32B-Instruct"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-Coder-32B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen2.5-Coder-32B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.03,
+        "output": 0.09
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "google/gemma-2-9b-it"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "google/gemma-2-9b-it"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.2
+      },
+      "context_length": 8192,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "google/gemma-2-9b-it"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "google/gemma-2-9b-it"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.4,
+        "output": 1.3
+      },
+      "context_length": 64000,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-V3"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.5,
+        "output": 1.5
+      },
+      "context_length": 163840,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-V3"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-V3"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 1.25,
+        "output": 1.25
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-V3"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "deepseek-ai/DeepSeek-V3"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "CohereLabs/aya-vision-8b"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.6
+      },
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.6
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "pricing": {
+        "input": 0.63,
+        "output": 1.8
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.55,
+        "output": 2.2
+      },
+      "context_length": 1000000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "MiniMaxAI/MiniMax-M1-80k"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "CohereLabs/aya-expanse-8b"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen2.5-Coder-7B-Instruct"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-Coder-7B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen2.5-Coder-7B-Instruct"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "CohereLabs/c4ai-command-a-03-2025"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-Coder-3B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.8,
+        "output": 0.8
+      },
+      "context_length": 32000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.25,
+        "output": 0.75
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B"
+    },
+    {
+      "provider": "cerebras",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.75,
+        "output": 0.99
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "pricing": {
+        "input": 0.7,
+        "output": 1.4
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.8
+      },
+      "context_length": 40960,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-235B-A22B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.6
+      },
+      "context_length": 40960,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-235B-A22B"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen3-235B-A22B"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.6
+      },
+      "context_length": 40960,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-235B-A22B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-235B-A22B"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "Qwen/Qwen3-235B-A22B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.13,
+        "output": 0.4
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "aaditya/Llama3-OpenBioLLM-70B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.04,
+        "output": 0.04
+      },
+      "context_length": 32000,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.3
+      },
+      "context_length": 16384,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "microsoft/phi-4"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.3,
+        "output": 0.3
+      },
+      "context_length": 64000,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.6,
+        "output": 1.8
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "nvidia/Llama-3_1-Nemotron-Ultra-253B-v1"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.51,
+        "output": 0.74
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "meta-llama/Meta-Llama-3-70B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "meta-llama/Meta-Llama-3-70B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.88,
+        "output": 0.88
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "meta-llama/Meta-Llama-3-70B-Instruct"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.59,
+        "output": 0.79
+      },
+      "context_length": 8192,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Meta-Llama-3-70B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Meta-Llama-3-70B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "meta-llama/Meta-Llama-3-70B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.05,
+        "output": 0.05
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Sao10K/L3-8B-Stheno-v3.2"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Sao10K/L3-8B-Stheno-v3.2"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Sao10K/L3-8B-Stheno-v3.2"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "CohereLabs/c4ai-command-r-plus"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.28,
+        "output": 1.1
+      },
+      "context_length": 123000,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "baidu/ERNIE-4.5-300B-A47B-Base-PT"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2.5-VL-32B-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "Qwen/Qwen2.5-VL-32B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.15,
+        "output": 0.15
+      },
+      "context_length": 64000,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-14B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-14B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-14B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-14B"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "CohereLabs/aya-expanse-32b"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "context_length": 120000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "baidu/ERNIE-4.5-0.3B-PT"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.38,
+        "output": 0.4
+      },
+      "context_length": 32000,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2.5-72B-Instruct"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.25,
+        "output": 0.75
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-72B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen2.5-72B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 1.2,
+        "output": 1.2
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-72B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2.5-72B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen2.5-72B-Instruct"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 1,
+        "output": 3
+      },
+      "context_length": 131072,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.1-405B-Instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-3.1-405B-Instruct"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "meta-llama/Llama-3.1-405B-Instruct"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "meta-llama/Llama-3.1-405B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.6
+      },
+      "context_length": 40960,
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen3-235B-A22B-FP8"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.42,
+        "output": 1.25
+      },
+      "context_length": 123000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "baidu/ERNIE-4.5-VL-424B-A47B-Base-PT"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.13,
+        "output": 0.4
+      },
+      "context_length": 32768,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2-VL-72B-Instruct"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.03,
+        "output": 0.09
+      },
+      "context_length": 32768,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-Coder-7B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen2.5-Coder-7B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen2.5-Coder-7B"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "CohereLabs/aya-vision-32b"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "CohereLabs/c4ai-command-r7b-arabic-02-2025"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.25,
+        "output": 0.75
+      },
+      "context_length": 32000,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-VL-72B-Instruct"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2.5-VL-72B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.14,
+        "output": 0.14
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "NousResearch/Hermes-2-Pro-Llama-3-8B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.5,
+        "output": 1.5
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/QwQ-32B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/QwQ-32B"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/QwQ-32B"
+    },
+    {
+      "provider": "nscale",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/QwQ-32B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/QwQ-32B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.17,
+        "output": 0.85
+      },
+      "context_length": 1048576,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.27,
+        "output": 0.85
+      },
+      "context_length": 1048576,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.6,
+        "output": 0.6
+      },
+      "context_length": 32768,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 1.48,
+        "output": 1.48
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Sao10K/L3-70B-Euryale-v2.1"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Sao10K/L3-70B-Euryale-v2.1"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Sao10K/L3-70B-Euryale-v2.1"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/QwQ-32B-Preview"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": true,
+      "id": "Qwen/QwQ-32B-Preview"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/QwQ-32B-Preview"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.7,
+        "output": 2.5
+      },
+      "context_length": 160000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "deepseek-ai/DeepSeek-Prover-V2-671B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "marin-community/marin-8b-instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.18000000000000002,
+        "output": 0.18000000000000002
+      },
+      "context_length": 4096,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "marin-community/marin-8b-instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "marin-community/marin-8b-instruct"
+    },
+    {
+      "provider": "fireworks-ai",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "SentientAGI/Dobby-Unhinged-Llama-3.3-70B"
+    },
+    {
+      "provider": "fireworks",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "context_length": 163840,
+      "supports_image_input": false,
+      "supports_function_calling": false,
+      "id": "SentientAGI/Dobby-Unhinged-Llama-3.3-70B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 1,
+        "output": 3
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "NousResearch/Hermes-3-Llama-3.1-405B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.62,
+        "output": 0.62
+      },
+      "context_length": 65535,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "alpindale/WizardLM-2-8x22B"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "alpindale/WizardLM-2-8x22B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "alpindale/WizardLM-2-8x22B"
+    },
+    {
+      "provider": "groq",
+      "status": "live",
+      "pricing": {
+        "input": 0.2,
+        "output": 0.2
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "meta-llama/Llama-Guard-4-12B"
+    },
+    {
+      "provider": "nebius",
+      "status": "live",
+      "pricing": {
+        "input": 0.13,
+        "output": 0.4
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "Qwen/Qwen2.5-32B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen2.5-32B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen2.5-32B-Instruct"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.9,
+        "output": 0.9
+      },
+      "context_length": 16000,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "dphn/dolphin-2.9.2-mixtral-8x22b"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Qwen/Qwen2-72B-Instruct"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.9,
+        "output": 0.9
+      },
+      "context_length": 32768,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Qwen/Qwen2-72B-Instruct"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Qwen/Qwen2-72B-Instruct"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "NousResearch/Hermes-3-Llama-3.1-70B"
+    },
+    {
+      "provider": "hyperbolic",
+      "status": "live",
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "NousResearch/Hermes-3-Llama-3.1-70B"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "NousResearch/Hermes-3-Llama-3.1-70B"
+    },
+    {
+      "provider": "novita",
+      "status": "live",
+      "pricing": {
+        "input": 0.05,
+        "output": 0.05
+      },
+      "context_length": 8192,
+      "supports_tools": false,
+      "supports_structured_output": false,
+      "id": "Sao10K/L3-8B-Lunaris-v1"
+    },
+    {
+      "provider": "featherless-ai",
+      "status": "live",
+      "id": "Sao10K/L3-8B-Lunaris-v1"
+    },
+    {
+      "provider": "featherless",
+      "status": "live",
+      "pricing": {
+        "input": 0.1,
+        "output": 0.1
+      },
+      "context_length": 16384,
+      "id": "Sao10K/L3-8B-Lunaris-v1"
+    },
+    {
+      "provider": "together",
+      "status": "live",
+      "pricing": {
+        "input": 0.88,
+        "output": 0.88
+      },
+      "context_length": 32768,
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "CohereLabs/c4ai-command-r-v01"
+    },
+    {
+      "provider": "sambanova",
+      "status": "live",
+      "pricing": {
+        "input": 0.6,
+        "output": 1.2
+      },
+      "context_length": 131072,
+      "supports_tools": false,
+      "supports_structured_output": true,
+      "id": "tokyotech-llm/Llama-3.3-Swallow-70B-Instruct-v0.4"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "CohereLabs/c4ai-command-r-08-2024"
+    },
+    {
+      "provider": "cohere",
+      "status": "live",
+      "supports_tools": true,
+      "supports_structured_output": false,
+      "id": "CohereLabs/c4ai-command-r7b-12-2024"
+    }
+  ]
+}

providers/aggregator.ts

@@ -0,0 +1,162 @@
+import { ProviderEntry, ProviderFetcher } from './types';
+import { getStaticPricing } from './static-pricing';
+import { NovitaFetcher } from './novita';
+import { SambaNovaFetcher } from './sambanova';
+import { GroqFetcher } from './groq';
+import { FeatherlessFetcher } from './featherless';
+import { TogetherFetcher } from './together';
+import { CohereFetcher } from './cohere';
+import { FireworksFetcher } from './fireworks';
+import { NebiusFetcher } from './nebius';
+import { HyperbolicFetcher } from './hyperbolic';
+import { CerebrasFetcher } from './cerebras';
+import { NScaleFetcher } from './nscale';
+
+export interface AggregatorConfig {
+  providers?: string[];  // Specific providers to fetch from
+  apiKeys?: {
+    [provider: string]: string;
+  };
+  concurrent?: number;   // Number of concurrent fetches
+  includeStaticPricing?: boolean;
+}
+
+export class ProviderAggregator {
+  private fetchers: Map<string, ProviderFetcher>;
+  private config: AggregatorConfig;
+
+  constructor(config: AggregatorConfig = {}) {
+    this.config = {
+      concurrent: 3,
+      includeStaticPricing: true,
+      ...config
+    };
+    
+    this.fetchers = new Map();
+    this.initializeFetchers();
+  }
+
+  private initializeFetchers() {
+    const apiKeys = this.config.apiKeys || {};
+    
+    // Initialize all available fetchers
+    this.fetchers.set('novita', new NovitaFetcher(apiKeys.novita));
+    this.fetchers.set('sambanova', new SambaNovaFetcher(apiKeys.sambanova));
+    this.fetchers.set('groq', new GroqFetcher(apiKeys.groq));
+    this.fetchers.set('featherless', new FeatherlessFetcher(apiKeys.featherless));
+    this.fetchers.set('together', new TogetherFetcher(apiKeys.together));
+    this.fetchers.set('cohere', new CohereFetcher(apiKeys.cohere));
+    this.fetchers.set('fireworks', new FireworksFetcher(apiKeys.fireworks));
+    this.fetchers.set('nebius', new NebiusFetcher(apiKeys.nebius));
+    this.fetchers.set('hyperbolic', new HyperbolicFetcher(apiKeys.hyperbolic));
+    this.fetchers.set('cerebras', new CerebrasFetcher(apiKeys.cerebras));
+    this.fetchers.set('nscale', new NScaleFetcher(apiKeys.nscale));
+  }
+
+  async fetchAllProviders(): Promise<Map<string, ProviderEntry[]>> {
+    const results = new Map<string, ProviderEntry[]>();
+    const providers = this.config.providers || Array.from(this.fetchers.keys());
+    
+    // Fetch in batches to respect rate limits
+    const batches = this.createBatches(providers, this.config.concurrent || 3);
+    
+    for (const batch of batches) {
+      const batchPromises = batch.map(async (provider) => {
+        const fetcher = this.fetchers.get(provider);
+        if (!fetcher) {
+          console.warn(`No fetcher found for provider: ${provider}`);
+          return { provider, entries: [] };
+        }
+        
+        try {
+          console.log(`Fetching models from ${provider}...`);
+          const entries = await fetcher.fetchModels();
+          
+          // Enrich with static pricing if needed
+          const enrichedEntries = this.enrichWithStaticPricing(provider, entries);
+          
+          return { provider, entries: enrichedEntries };
+        } catch (error) {
+          console.error(`Failed to fetch from ${provider}:`, error);
+          return { provider, entries: [] };
+        }
+      });
+      
+      const batchResults = await Promise.all(batchPromises);
+      for (const { provider, entries } of batchResults) {
+        results.set(provider, entries);
+      }
+    }
+    
+    return results;
+  }
+
+  async fetchProvider(provider: string): Promise<ProviderEntry[]> {
+    const fetcher = this.fetchers.get(provider);
+    if (!fetcher) {
+      throw new Error(`No fetcher found for provider: ${provider}`);
+    }
+    
+    const entries = await fetcher.fetchModels();
+    return this.enrichWithStaticPricing(provider, entries);
+  }
+
+  private enrichWithStaticPricing(provider: string, entries: ProviderEntry[]): ProviderEntry[] {
+    if (!this.config.includeStaticPricing) {
+      return entries;
+    }
+    
+    return entries.map(entry => {
+      // Only add static pricing if the entry doesn't already have pricing
+      if (!entry.pricing) {
+        const modelId = this.extractModelId(entry);
+        const staticPrice = getStaticPricing(provider, modelId);
+        if (staticPrice) {
+          return {
+            ...entry,
+            pricing: staticPrice
+          };
+        }
+      }
+      return entry;
+    });
+  }
+
+  private extractModelId(entry: ProviderEntry): string {
+    // Extract model ID from various possible fields
+    // This is a simplified version - in production you'd need provider-specific logic
+    return (entry as any).id || (entry as any).model_id || 'unknown';
+  }
+
+  private createBatches<T>(items: T[], batchSize: number): T[][] {
+    const batches: T[][] = [];
+    for (let i = 0; i < items.length; i += batchSize) {
+      batches.push(items.slice(i, i + batchSize));
+    }
+    return batches;
+  }
+
+  // Aggregate all provider data into a single array
+  async aggregateAll(): Promise<ProviderEntry[]> {
+    const providerMap = await this.fetchAllProviders();
+    const allEntries: ProviderEntry[] = [];
+    
+    for (const [provider, entries] of providerMap) {
+      allEntries.push(...entries);
+    }
+    
+    return allEntries;
+  }
+
+  // Get a summary of available models per provider
+  async getSummary(): Promise<{ [provider: string]: number }> {
+    const providerMap = await this.fetchAllProviders();
+    const summary: { [provider: string]: number } = {};
+    
+    for (const [provider, entries] of providerMap) {
+      summary[provider] = entries.length;
+    }
+    
+    return summary;
+  }
+}

providers/base.ts

@@ -0,0 +1,200 @@
+import { ProviderEntry, ProviderFetcher, RateLimitConfig } from "./types";
+
+export abstract class BaseProviderFetcher implements ProviderFetcher {
+  abstract name: string;
+  protected apiKey?: string;
+  protected baseUrl: string;
+  protected rateLimitConfig: RateLimitConfig;
+
+  private lastRequestTime: number = 0;
+  private requestCount: number = 0;
+  private requestWindowStart: number = Date.now();
+
+  constructor(
+    baseUrl: string,
+    apiKey?: string,
+    rateLimitConfig: RateLimitConfig = {}
+  ) {
+    this.baseUrl = baseUrl;
+    this.apiKey = apiKey;
+    this.rateLimitConfig = {
+      requestsPerMinute: rateLimitConfig.requestsPerMinute || 60,
+      retryAttempts: rateLimitConfig.retryAttempts || 3,
+      initialBackoffMs: rateLimitConfig.initialBackoffMs || 1000,
+      ...rateLimitConfig,
+    };
+  }
+
+  abstract fetchModels(): Promise<ProviderEntry[]>;
+
+  protected async fetchWithRetry<T>(
+    url: string,
+    options: RequestInit = {},
+    retries: number = this.rateLimitConfig.retryAttempts || 3
+  ): Promise<T> {
+    // Apply rate limiting
+    await this.enforceRateLimit();
+
+    for (let i = 0; i < retries; i++) {
+      try {
+        const headers: HeadersInit = {
+          "Content-Type": "application/json",
+          ...(options.headers || {}),
+        };
+
+        if (this.apiKey) {
+          headers["Authorization"] = `Bearer ${this.apiKey}`;
+        }
+
+        const response = await fetch(url, {
+          ...options,
+          headers,
+        });
+
+        if (response.ok) {
+          return (await response.json()) as T;
+        }
+
+        // Handle rate limit errors
+        if (response.status === 429) {
+          const retryAfter = response.headers.get("Retry-After");
+          const waitTime = retryAfter
+            ? parseInt(retryAfter) * 1000
+            : Math.pow(2, i) * (this.rateLimitConfig.initialBackoffMs || 1000);
+
+          console.log(`Rate limited by ${this.name}, waiting ${waitTime}ms...`);
+          await this.sleep(waitTime);
+          continue;
+        }
+
+        // Handle other errors
+        const errorBody = await response.text();
+        throw new Error(
+          `HTTP ${response.status}: ${response.statusText} - ${errorBody}`
+        );
+      } catch (error) {
+        if (i === retries - 1) {
+          console.error(
+            `Failed to fetch from ${this.name} after ${retries} attempts:`,
+            error
+          );
+          throw error;
+        }
+
+        // Exponential backoff for other errors
+        const waitTime =
+          Math.pow(2, i) * (this.rateLimitConfig.initialBackoffMs || 1000);
+        console.log(`Retrying ${this.name} request in ${waitTime}ms...`);
+        await this.sleep(waitTime);
+      }
+    }
+
+    throw new Error(
+      `Failed to fetch from ${this.name} after ${retries} attempts`
+    );
+  }
+
+  private async enforceRateLimit(): Promise<void> {
+    const now = Date.now();
+    const windowDuration = 60000; // 1 minute in milliseconds
+
+    // Reset window if needed
+    if (now - this.requestWindowStart >= windowDuration) {
+      this.requestCount = 0;
+      this.requestWindowStart = now;
+    }
+
+    // Check if we've hit the rate limit
+    if (this.requestCount >= (this.rateLimitConfig.requestsPerMinute || 60)) {
+      const waitTime = windowDuration - (now - this.requestWindowStart);
+      console.log(
+        `Rate limit reached for ${this.name}, waiting ${waitTime}ms...`
+      );
+      await this.sleep(waitTime);
+
+      // Reset after waiting
+      this.requestCount = 0;
+      this.requestWindowStart = Date.now();
+    }
+
+    // Ensure minimum time between requests (100ms default)
+    const timeSinceLastRequest = now - this.lastRequestTime;
+    const minInterval = 100;
+    if (timeSinceLastRequest < minInterval) {
+      await this.sleep(minInterval - timeSinceLastRequest);
+    }
+
+    this.requestCount++;
+    this.lastRequestTime = Date.now();
+  }
+
+  protected sleep(ms: number): Promise<void> {
+    return new Promise((resolve) => setTimeout(resolve, ms));
+  }
+
+  // Helper method to convert various price formats to $ per 1M tokens
+  protected normalizePricing(
+    input: number | string,
+    output: number | string,
+    unit: "per_token" | "per_million" | "cents_per_million" = "per_million"
+  ): { input: number; output: number } {
+    let inputPrice = typeof input === "string" ? parseFloat(input) : input;
+    let outputPrice = typeof output === "string" ? parseFloat(output) : output;
+
+    switch (unit) {
+      case "per_token":
+        // Convert from $ per token to $ per million tokens
+        inputPrice = inputPrice * 1_000_000;
+        outputPrice = outputPrice * 1_000_000;
+        break;
+      case "cents_per_million":
+        // Convert from cents per million to $ per million
+        inputPrice = inputPrice / 100;
+        outputPrice = outputPrice / 100;
+        break;
+      case "per_million":
+        // Already in the correct format
+        break;
+    }
+
+    return {
+      input: inputPrice,
+      output: outputPrice,
+    };
+  }
+
+  // Helper to parse supported parameters from various formats
+  protected parseSupportedParameters(
+    params: string[] | object
+  ): Partial<ProviderEntry> {
+    const result: Partial<ProviderEntry> = {};
+    const paramList = Array.isArray(params) ? params : Object.keys(params);
+
+    const paramMapping: { [key: string]: keyof ProviderEntry } = {
+      temperature: "supports_temperature",
+      top_p: "supports_top_p",
+      top_k: "supports_top_k",
+      max_tokens: "supports_max_tokens",
+      stop: "supports_stop_sequences",
+      seed: "supports_seed",
+      frequency_penalty: "supports_frequency_penalty",
+      presence_penalty: "supports_presence_penalty",
+      repetition_penalty: "supports_repetition_penalty",
+      min_p: "supports_min_p",
+      logit_bias: "supports_logit_bias",
+      logprobs: "supports_logprobs",
+      top_logprobs: "supports_top_logprobs",
+      stream: "supports_streaming",
+    };
+
+    for (const param of paramList) {
+      const mappedKey = paramMapping[param];
+      if (mappedKey) {
+        result[mappedKey] = true;
+      }
+    }
+
+    result.supported_parameters = paramList;
+    return result;
+  }
+}

providers/cerebras.ts

@@ -0,0 +1,17 @@
+import type { ProviderEntry, ProviderFetcher } from './types';
+import { BaseProviderFetcher } from './base';
+
+export class CerebrasFetcher extends BaseProviderFetcher implements ProviderFetcher {
+  name = 'cerebras';
+  
+  constructor(apiKey?: string) {
+    super('https://api.cerebras.ai/v1', apiKey);
+  }
+  
+  async fetchModels(): Promise<ProviderEntry[]> {
+    // Cerebras doesn't provide detailed model information via their API
+    // Data will come from HuggingFace router API
+    console.log('Cerebras API limited - using HuggingFace router data');
+    return [];
+  }
+}

providers/cohere.ts

@@ -0,0 +1,102 @@
+import { BaseProviderFetcher } from './base';
+import type { ProviderEntry, CohereModel } from './types';
+
+export class CohereFetcher extends BaseProviderFetcher {
+  name = 'cohere';
+
+  constructor(apiKey?: string) {
+    super('https://api.cohere.ai', apiKey, {
+      requestsPerMinute: 60  // Conservative default
+    });
+  }
+
+  async fetchModels(): Promise<ProviderEntry[]> {
+    try {
+      // Fetch all models
+      const response = await this.fetchWithRetry<{ models: CohereModel[] }>(
+        `${this.baseUrl}/v1/models`
+      );
+
+      // Optionally filter by endpoint type
+      const chatModels = response.models.filter(model => 
+        model.endpoints.includes('chat') || model.endpoints.includes('generate')
+      );
+
+      return chatModels.map(model => this.mapModelToProviderEntry(model));
+    } catch (error) {
+      console.error(`Failed to fetch Cohere models: ${error}`);
+      return [];
+    }
+  }
+
+  async fetchModel(modelName: string): Promise<ProviderEntry | null> {
+    try {
+      const response = await this.fetchWithRetry<CohereModel>(
+        `${this.baseUrl}/v1/models/${encodeURIComponent(modelName)}`
+      );
+
+      return this.mapModelToProviderEntry(response);
+    } catch (error) {
+      console.error(`Failed to fetch Cohere model ${modelName}: ${error}`);
+      return null;
+    }
+  }
+
+  private mapModelToProviderEntry(model: CohereModel): ProviderEntry {
+    const entry: ProviderEntry = {
+      provider: this.name,
+      context_length: model.context_length,
+      status: model.is_deprecated ? 'deprecated' : 'live',
+      supports_image_input: model.supports_vision
+    };
+
+    // Map features to capability flags
+    const featureMapping = this.mapFeatures(model.features);
+    Object.assign(entry, featureMapping);
+
+    // Map endpoints to capabilities
+    const endpointCapabilities = this.mapEndpoints(model.endpoints);
+    Object.assign(entry, endpointCapabilities);
+
+    // Set supported parameters based on features
+    entry.supported_parameters = model.features;
+
+    return entry;
+  }
+
+  private mapFeatures(features: string[]): Partial<ProviderEntry> {
+    const result: Partial<ProviderEntry> = {};
+
+    // Feature mapping based on the spec
+    const featureMap: { [key: string]: (keyof ProviderEntry)[] } = {
+      'tools': ['supports_tools'],
+      'strict_tools': ['supports_function_calling'],
+      'json_mode': ['supports_structured_output'],
+      'json_schema': ['supports_structured_output', 'supports_response_format'],
+      'logprobs': ['supports_logprobs']
+    };
+
+    for (const feature of features) {
+      const mappedKeys = featureMap[feature];
+      if (mappedKeys) {
+        for (const key of mappedKeys) {
+          (result[key] as any) = true;
+        }
+      }
+    }
+
+
+    return result;
+  }
+
+  private mapEndpoints(endpoints: string[]): Partial<ProviderEntry> {
+    const result: Partial<ProviderEntry> = {};
+
+    // If the model supports chat or generate endpoints, it's a text generation model
+    if (endpoints.includes('chat') || endpoints.includes('generate')) {
+      result.model_type = 'chat';
+    }
+
+    return result;
+  }
+}

providers/featherless.ts

@@ -0,0 +1,40 @@
+import { BaseProviderFetcher } from "./base";
+import { ProviderEntry, FeatherlessModel } from "./types";
+
+export class FeatherlessFetcher extends BaseProviderFetcher {
+  name = "featherless";
+
+  constructor(apiKey?: string) {
+    super("https://api.featherless.ai", apiKey, {
+      requestsPerMinute: 60, // Conservative default
+    });
+  }
+
+  async fetchModels(): Promise<ProviderEntry[]> {
+    try {
+      const response = await this.fetchWithRetry<{ data: FeatherlessModel[] }>(
+        `${this.baseUrl}/v1/models`
+      );
+
+      return response.data.map((model) => this.mapModelToProviderEntry(model));
+    } catch (error) {
+      console.error(`Failed to fetch Featherless models: ${error}`);
+      return [];
+    }
+  }
+
+  private mapModelToProviderEntry(model: FeatherlessModel): ProviderEntry {
+    const entry: ProviderEntry = {
+      provider: this.name,
+      context_length: model.context_length,
+      max_completion_tokens: model.max_completion_tokens,
+      status: model.available_on_current_plan ? "live" : "offline",
+      owned_by: model.owned_by,
+      model_class: model.model_class,
+      is_gated: model.is_gated,
+    };
+
+
+    return entry;
+  }
+}

providers/fireworks.ts

@@ -0,0 +1,130 @@
+import { BaseProviderFetcher } from './base';
+import { ProviderEntry, FireworksModel, FireworksDetailedModel } from './types';
+
+export class FireworksFetcher extends BaseProviderFetcher {
+  name = 'fireworks';
+
+  constructor(apiKey?: string) {
+    super('https://api.fireworks.ai', apiKey, {
+      requestsPerMinute: 60  // Conservative default
+    });
+  }
+
+  async fetchModels(): Promise<ProviderEntry[]> {
+    try {
+      const response = await this.fetchWithRetry<{ data: FireworksModel[] }>(
+        `${this.baseUrl}/inference/v1/models`
+      );
+
+      // Map basic model data
+      const basicEntries = response.data.map(model => this.mapBasicModelToProviderEntry(model));
+
+      // Optionally enrich with detailed data for important models
+      // This can be done selectively to avoid too many API calls
+      const enrichedEntries = await this.enrichModels(basicEntries, response.data);
+
+      return enrichedEntries;
+    } catch (error) {
+      console.error(`Failed to fetch Fireworks models: ${error}`);
+      return [];
+    }
+  }
+
+  private async enrichModels(
+    basicEntries: ProviderEntry[], 
+    models: FireworksModel[]
+  ): Promise<ProviderEntry[]> {
+    // For now, we'll return basic entries
+    // In production, you might want to selectively enrich important models
+    // to avoid hitting rate limits
+    return basicEntries;
+  }
+
+  async fetchDetailedModel(accountId: string, modelId: string): Promise<ProviderEntry | null> {
+    try {
+      const response = await this.fetchWithRetry<FireworksDetailedModel>(
+        `${this.baseUrl}/v1/accounts/${accountId}/models/${modelId}`
+      );
+
+      return this.mapDetailedModelToProviderEntry(response);
+    } catch (error) {
+      console.error(`Failed to fetch detailed Fireworks model ${modelId}: ${error}`);
+      return null;
+    }
+  }
+
+  private mapBasicModelToProviderEntry(model: FireworksModel): ProviderEntry {
+    const entry: ProviderEntry = {
+      provider: this.name,
+      context_length: model.context_length,
+      owned_by: model.owned_by,
+      supports_image_input: model.supports_image_input,
+      supports_tools: model.supports_tools,
+      supports_function_calling: model.supports_tools
+    };
+
+    // Set model type based on chat support
+    if (model.supports_chat) {
+      entry.model_type = 'chat';
+    }
+
+    return entry;
+  }
+
+  private mapDetailedModelToProviderEntry(model: FireworksDetailedModel): ProviderEntry {
+    const entry: ProviderEntry = {
+      provider: this.name,
+      context_length: model.contextLength,
+      status: model.state === 'READY' ? 'live' : 'offline',
+      description: model.description,
+      quantization: model.baseModelDetails.defaultPrecision,
+      supports_image_input: model.supportsImageInput,
+      supports_tools: model.supportsTools,
+      supports_function_calling: model.supportsTools
+    };
+
+    // Check deprecation
+    if (model.deprecationDate) {
+      entry.status = 'deprecated';
+      entry.deprecated_at = model.deprecationDate;
+    }
+
+    // Parse parameter count if available
+    if (model.baseModelDetails.parameterCount) {
+      // Store as metadata - you might want to parse this into a number
+      entry.owned_by = model.displayName;
+    }
+
+    // Parse supported parameters from defaultSamplingParams
+    if (model.defaultSamplingParams) {
+      const paramCapabilities = this.parseSupportedParameters(model.defaultSamplingParams);
+      Object.assign(entry, paramCapabilities);
+    }
+
+    // Additional capabilities from model details
+    if (model.supportsLora) {
+      // Custom capability - not in standard ProviderEntry but could be added
+      // entry.supports_lora = true;
+    }
+
+    // Map supported precisions
+    if (model.supportedPrecisions && model.supportedPrecisions.length > 0) {
+      // Could store as metadata or custom field
+    }
+
+    return entry;
+  }
+
+  // Helper to extract model ID parts from Fireworks model ID format
+  private parseModelId(id: string): { accountId: string; modelId: string } | null {
+    // Format: "accounts/fireworks/models/qwen3-235b-a22b-thinking-2507"
+    const match = id.match(/accounts\/([^\/]+)\/models\/([^\/]+)/);
+    if (match) {
+      return {
+        accountId: match[1],
+        modelId: match[2]
+      };
+    }
+    return null;
+  }
+}

providers/groq.ts

@@ -0,0 +1,68 @@
+import { BaseProviderFetcher } from "./base";
+import { ProviderEntry, GroqModel } from "./types";
+
+export class GroqFetcher extends BaseProviderFetcher {
+  name = "groq";
+
+  constructor(apiKey?: string) {
+    super("https://api.groq.com", apiKey, {
+      requestsPerMinute: 100, // Groq rate limit from spec
+    });
+  }
+
+  async fetchModels(): Promise<ProviderEntry[]> {
+    try {
+      const response = await this.fetchWithRetry<{ data: GroqModel[] }>(
+        `${this.baseUrl}/openai/v1/models`
+      );
+
+      return response.data.map((model) => this.mapModelToProviderEntry(model));
+    } catch (error) {
+      console.error(`Failed to fetch Groq models: ${error}`);
+      return [];
+    }
+  }
+
+  async fetchModel(modelId: string): Promise<ProviderEntry | null> {
+    try {
+      const response = await this.fetchWithRetry<GroqModel>(
+        `${this.baseUrl}/openai/v1/models/${encodeURIComponent(modelId)}`
+      );
+
+      return this.mapModelToProviderEntry(response);
+    } catch (error) {
+      console.error(`Failed to fetch Groq model ${modelId}: ${error}`);
+      return null;
+    }
+  }
+
+  private mapModelToProviderEntry(model: GroqModel): ProviderEntry {
+    const entry: ProviderEntry = {
+      provider: this.name,
+      context_length: model.context_window,
+      max_completion_tokens: model.max_completion_tokens,
+      status: model.active ? "live" : "offline",
+      owned_by: model.owned_by,
+    };
+
+    // Store the model ID for matching
+    (entry as any).id = model.id;
+
+    // Add static pricing from Groq's website if not provided by API
+    if (!entry.pricing) {
+      const staticPricing = this.getStaticPricing(model.id);
+      if (staticPricing) {
+        entry.pricing = staticPricing;
+      }
+    }
+
+
+    return entry;
+  }
+
+  private getStaticPricing(modelId: string): { input: number; output: number } | null {
+    // Import static pricing data
+    const { getStaticPricing } = require('./static-pricing');
+    return getStaticPricing('groq', modelId);
+  }
+}

providers/huggingface-router.ts

@@ -0,0 +1,149 @@
+import type { ProviderEntry, ProviderFetcher } from './types';
+
+interface HFRouterModel {
+  id: string;
+  object: string;
+  created: number;
+  owned_by: string;
+  providers?: HFRouterProvider[];
+}
+
+interface HFRouterProvider {
+  provider: string;
+  status?: "live" | "offline" | "staging" | "deprecated";
+  context_length?: number;
+  pricing?: {
+    input: number;   // cents per million tokens
+    output: number;  // cents per million tokens
+  };
+  supports_tools?: boolean;
+  supports_structured_output?: boolean;
+}
+
+export class HuggingFaceRouterFetcher implements ProviderFetcher {
+  name = 'huggingface-router';
+  
+  async fetchModels(): Promise<ProviderEntry[]> {
+    try {
+      const response = await fetch('https://router.huggingface.co/v1/models');
+      if (!response.ok) {
+        throw new Error(`HTTP ${response.status}: ${response.statusText}`);
+      }
+      
+      const data = await response.json() as { data: HFRouterModel[] };
+      return this.normalizeModels(data.data);
+    } catch (error) {
+      console.error('Failed to fetch HuggingFace router models:', error);
+      throw error;
+    }
+  }
+  
+  private normalizeModels(models: HFRouterModel[]): ProviderEntry[] {
+    const entries: ProviderEntry[] = [];
+    
+    for (const model of models) {
+      if (!model.providers) continue;
+      
+      for (const provider of model.providers) {
+        const entry: ProviderEntry = {
+          provider: this.normalizeProviderName(provider.provider),
+          model_id: model.id,
+          owned_by: model.owned_by,
+          created: model.created,
+        };
+        
+        // Set status
+        if (provider.status) {
+          entry.status = provider.status === "staging" ? "offline" : provider.status;
+        }
+        
+        // Convert pricing from cents to dollars per million tokens
+        if (provider.pricing) {
+          entry.pricing = {
+            input: provider.pricing.input / 100,   // cents to dollars
+            output: provider.pricing.output / 100, // cents to dollars
+          };
+        }
+        
+        // Copy context length
+        if (provider.context_length) {
+          entry.context_length = provider.context_length;
+        }
+        
+        // Copy capability flags
+        if (provider.supports_tools !== undefined) {
+          entry.supports_tools = provider.supports_tools;
+        }
+        
+        if (provider.supports_structured_output !== undefined) {
+          entry.supports_structured_output = provider.supports_structured_output;
+        }
+        
+        entries.push(entry);
+      }
+    }
+    
+    return entries;
+  }
+  
+  private normalizeProviderName(providerName: string): string {
+    // Map HF router provider names to our standard names
+    const providerMap: Record<string, string> = {
+      'featherless-ai': 'featherless',
+      'fireworks-ai': 'fireworks',
+      'hf-inference': 'huggingface',
+      // Keep others as-is
+    };
+    
+    return providerMap[providerName] || providerName;
+  }
+}
+
+// Helper function to extract HF router data from a model
+export function extractHFRouterData(model: any): Map<string, ProviderEntry> {
+  const providerMap = new Map<string, ProviderEntry>();
+  
+  if (!model.providers || !Array.isArray(model.providers)) {
+    return providerMap;
+  }
+  
+  for (const provider of model.providers) {
+    if (!provider.provider) continue;
+    
+    const entry: ProviderEntry = {
+      provider: provider.provider,
+    };
+    
+    // Set status
+    if (provider.status) {
+      entry.status = provider.status === "staging" ? "offline" : provider.status;
+    }
+    
+    // Convert pricing from cents to dollars if needed
+    if (provider.pricing) {
+      // Check if pricing is already in dollars (values < 100 likely dollars)
+      const needsConversion = provider.pricing.input >= 100 || provider.pricing.output >= 100;
+      entry.pricing = {
+        input: needsConversion ? provider.pricing.input / 100 : provider.pricing.input,
+        output: needsConversion ? provider.pricing.output / 100 : provider.pricing.output,
+      };
+    }
+    
+    // Copy other fields
+    if (provider.context_length) {
+      entry.context_length = provider.context_length;
+    }
+    
+    if (provider.supports_tools !== undefined) {
+      entry.supports_tools = provider.supports_tools;
+    }
+    
+    if (provider.supports_structured_output !== undefined) {
+      entry.supports_structured_output = provider.supports_structured_output;
+    }
+    
+    providerMap.set(provider.provider, entry);
+  }
+  
+  return providerMap;
+}

providers/hyperbolic.ts

@@ -0,0 +1,17 @@
+import type { ProviderEntry, ProviderFetcher } from './types';
+import { BaseProviderFetcher } from './base';
+
+export class HyperbolicFetcher extends BaseProviderFetcher implements ProviderFetcher {
+  name = 'hyperbolic';
+  
+  constructor(apiKey?: string) {
+    super('https://api.hyperbolic.ai/v1', apiKey);
+  }
+  
+  async fetchModels(): Promise<ProviderEntry[]> {
+    // Hyperbolic doesn't provide a public API for model listing
+    // Data will come from HuggingFace router API
+    console.log('Hyperbolic API not available - using HuggingFace router data');
+    return [];
+  }
+}

providers/index.ts

@@ -0,0 +1,12 @@
+// Main export file for provider module
+export * from './types';
+export * from './base';
+export * from './novita';
+export * from './sambanova';
+export * from './groq';
+export * from './featherless';
+export * from './together';
+export * from './cohere';
+export * from './fireworks';
+export * from './static-pricing';
+export * from './aggregator';

providers/nebius.ts

@@ -0,0 +1,17 @@
+import type { ProviderEntry, ProviderFetcher } from './types';
+import { BaseProviderFetcher } from './base';
+
+export class NebiusFetcher extends BaseProviderFetcher implements ProviderFetcher {
+  name = 'nebius';
+  
+  constructor(apiKey?: string) {
+    super('https://api.nebius.ai/v1', apiKey);
+  }
+  
+  async fetchModels(): Promise<ProviderEntry[]> {
+    // Nebius doesn't provide a public API for model listing
+    // Data will come from HuggingFace router API
+    console.log('Nebius API not available - using HuggingFace router data');
+    return [];
+  }
+}

providers/novita.ts

@@ -0,0 +1,96 @@
+import { BaseProviderFetcher } from "./base";
+import { ProviderEntry, NovitaModel } from "./types";
+
+export class NovitaFetcher extends BaseProviderFetcher {
+  name = "novita";
+
+  constructor(apiKey?: string) {
+    super("https://api.novita.ai", apiKey, {
+      requestsPerMinute: 60, // Conservative default
+    });
+  }
+
+  async fetchModels(): Promise<ProviderEntry[]> {
+    try {
+      const response = await this.fetchWithRetry<{ data: NovitaModel[] }>(
+        `${this.baseUrl}/v3/openai/models`
+      );
+
+      return response.data.map((model) => this.mapModelToProviderEntry(model));
+    } catch (error) {
+      console.error(`Failed to fetch Novita models: ${error}`);
+      return [];
+    }
+  }
+
+  private mapModelToProviderEntry(model: NovitaModel): ProviderEntry {
+    const entry: ProviderEntry = {
+      provider: this.name,
+      context_length: model.context_size,
+      max_completion_tokens: model.max_output_tokens,
+      pricing: this.normalizePricing(
+        model.input_token_price_per_m,
+        model.output_token_price_per_m,
+        "cents_per_million"
+      ),
+      description: model.description,
+      model_type: model.model_type,
+      status: model.status === 1 ? "live" : "offline",
+    };
+
+    // Store the model ID for matching
+    (entry as any).id = model.id;
+
+    // Map features to capability flags if features exist
+    if (model.features && Array.isArray(model.features)) {
+      const featureMapping = this.mapFeatures(model.features);
+      Object.assign(entry, featureMapping);
+    }
+
+    // Add additional metadata
+    if (model.display_name) {
+      entry.owned_by = model.owned_by || "unknown";
+    }
+
+
+    return entry;
+  }
+
+  private mapFeatures(features: string[]): Partial<ProviderEntry> {
+    const result: Partial<ProviderEntry> = {};
+
+    // Feature mapping based on the spec
+    const featureMap: { [key: string]: (keyof ProviderEntry)[] } = {
+      "function-calling": ["supports_tools", "supports_function_calling"],
+      "structured-outputs": [
+        "supports_structured_output",
+        "supports_response_format",
+      ],
+    };
+
+    for (const feature of features || []) {
+      const mappedKeys = featureMap[feature];
+      if (mappedKeys) {
+        for (const key of mappedKeys) {
+          result[key] = true;
+        }
+      }
+    }
+
+    return result;
+  }
+
+  // Optional: Fetch a single model with potentially more details
+  async fetchModel(modelId: string): Promise<ProviderEntry | null> {
+    try {
+      const response = await this.fetchWithRetry<NovitaModel>(
+        `${this.baseUrl}/v3/openai/models/${encodeURIComponent(modelId)}`
+      );
+
+      return this.mapModelToProviderEntry(response);
+    } catch (error) {
+      console.error(`Failed to fetch Novita model ${modelId}: ${error}`);
+      return null;
+    }
+  }
+}

providers/nscale.ts

@@ -0,0 +1,17 @@
+import type { ProviderEntry, ProviderFetcher } from './types';
+import { BaseProviderFetcher } from './base';
+
+export class NScaleFetcher extends BaseProviderFetcher implements ProviderFetcher {
+  name = 'nscale';
+  
+  constructor(apiKey?: string) {
+    super('https://api.nscale.ai/v1', apiKey);
+  }
+  
+  async fetchModels(): Promise<ProviderEntry[]> {
+    // NScale doesn't provide a public API for model listing
+    // Data will come from HuggingFace router API
+    console.log('NScale API not available - using HuggingFace router data');
+    return [];
+  }
+}

providers/sambanova.ts

@@ -0,0 +1,45 @@
+import { BaseProviderFetcher } from "./base";
+import { ProviderEntry, SambaNovaModel } from "./types";
+
+export class SambaNovaFetcher extends BaseProviderFetcher {
+  name = "sambanova";
+
+  constructor(apiKey?: string) {
+    super("https://api.sambanova.ai", apiKey, {
+      requestsPerMinute: 60, // Conservative default
+    });
+  }
+
+  async fetchModels(): Promise<ProviderEntry[]> {
+    try {
+      const response = await this.fetchWithRetry<{ data: SambaNovaModel[] }>(
+        `${this.baseUrl}/v1/models`
+      );
+
+      return response.data.map((model) => this.mapModelToProviderEntry(model));
+    } catch (error) {
+      console.error(`Failed to fetch SambaNova models: ${error}`);
+      return [];
+    }
+  }
+
+  private mapModelToProviderEntry(model: SambaNovaModel): ProviderEntry {
+    const entry: ProviderEntry = {
+      provider: this.name,
+      context_length: model.context_length,
+      max_completion_tokens: model.max_completion_tokens,
+      pricing: this.normalizePricing(
+        model.pricing.prompt,
+        model.pricing.completion,
+        "per_token"
+      ),
+      owned_by: model.owned_by,
+    };
+
+    // Store the model ID for matching
+    (entry as any).id = model.id;
+
+
+    return entry;
+  }
+}

providers/static-pricing.ts

@@ -0,0 +1,112 @@
+import { StaticPricing } from './types';
+
+// Static pricing data for providers without API pricing endpoints
+// Prices are in $ per 1M tokens
+// Last updated: January 2025
+export const staticPricing: StaticPricing = {
+  groq: {
+    // Groq pricing from their website
+    'llama-3.3-70b-versatile': { input: 0.59, output: 0.79 },
+    'llama-3.1-70b-versatile': { input: 0.59, output: 0.79 },
+    'llama-3.1-8b-instant': { input: 0.05, output: 0.08 },
+    'llama-3.2-1b-preview': { input: 0.04, output: 0.04 },
+    'llama-3.2-3b-preview': { input: 0.06, output: 0.06 },
+    'llama-3.2-11b-vision-preview': { input: 0.18, output: 0.18 },
+    'llama-3.2-90b-vision-preview': { input: 0.90, output: 0.90 },
+    'llama3-70b-8192': { input: 0.59, output: 0.79 },
+    'llama3-8b-8192': { input: 0.05, output: 0.08 },
+    'mixtral-8x7b-32768': { input: 0.24, output: 0.24 },
+    'gemma-7b-it': { input: 0.07, output: 0.07 },
+    'gemma2-9b-it': { input: 0.20, output: 0.20 }
+  },
+  
+  featherless: {
+    // Featherless pricing - typically uses pay-per-request model
+    // Converting to per-million-token estimates based on average usage
+    'default': { input: 0.10, output: 0.10 }  // Default pricing for all models
+  },
+  
+  cohere: {
+    // Cohere pricing from their website
+    'command-r-plus': { input: 2.50, output: 10.00 },
+    'command-r': { input: 0.15, output: 0.60 },
+    'command': { input: 0.50, output: 1.50 },
+    'command-light': { input: 0.15, output: 0.60 },
+    'c4ai-aya-expanse-8b': { input: 0.15, output: 0.60 },
+    'c4ai-aya-expanse-32b': { input: 0.50, output: 2.00 }
+  },
+  
+  fireworks: {
+    // Fireworks pricing from their documentation
+    'qwen2.5-coder-32b-instruct': { input: 0.25, output: 0.25 },
+    'qwen2.5-72b-instruct': { input: 0.50, output: 0.50 },
+    'llama-v3p3-70b-instruct': { input: 0.50, output: 0.50 },
+    'llama-v3p2-11b-vision-instruct': { input: 0.20, output: 0.20 },
+    'llama-v3p2-90b-vision-instruct': { input: 1.00, output: 1.00 },
+    'llama-v3p1-405b-instruct': { input: 3.00, output: 3.00 },
+    'llama-v3p1-70b-instruct': { input: 0.50, output: 0.50 },
+    'llama-v3p1-8b-instruct': { input: 0.10, output: 0.10 },
+    'mixtral-8x7b-instruct': { input: 0.50, output: 0.50 },
+    'mixtral-8x22b-instruct': { input: 0.90, output: 0.90 },
+    'deepseek-v3': { input: 0.30, output: 0.30 },
+    'mythomax-l2-13b': { input: 0.10, output: 0.10 }
+  },
+  
+  cerebras: {
+    // Cerebras pricing - very competitive
+    'llama3.1-8b': { input: 0.10, output: 0.10 },
+    'llama3.1-70b': { input: 0.60, output: 0.60 }
+  },
+  
+  nebius: {
+    // Nebius pricing estimates
+    'llama-3.1-70b-instruct': { input: 0.50, output: 0.50 },
+    'llama-3.1-8b-instruct': { input: 0.10, output: 0.10 },
+    'llama-3.1-405b-instruct': { input: 2.50, output: 2.50 },
+    'mistral-7b-instruct': { input: 0.10, output: 0.10 }
+  },
+  
+  lambdalabs: {
+    // Lambda Labs pricing - typically hourly GPU pricing
+    // These are estimates based on typical usage patterns
+    'hermes-3-llama-3.1-405b-fp8': { input: 3.00, output: 3.00 },
+    'hermes-3-llama-3.1-70b-fp8': { input: 0.50, output: 0.50 }
+  },
+  
+  lepton: {
+    // Lepton AI pricing
+    'llama3.1-8b': { input: 0.10, output: 0.10 },
+    'llama3.1-70b': { input: 0.50, output: 0.50 },
+    'llama3.1-405b': { input: 2.50, output: 2.50 },
+    'qwen2.5-72b': { input: 0.50, output: 0.50 },
+    'mixtral-8x7b': { input: 0.30, output: 0.30 }
+  },
+  
+  octoai: {
+    // OctoAI pricing
+    'meta-llama-3.1-8b-instruct': { input: 0.05, output: 0.10 },
+    'meta-llama-3.1-70b-instruct': { input: 0.50, output: 0.50 },
+    'meta-llama-3.1-405b-instruct': { input: 2.50, output: 2.50 },
+    'qwen2.5-72b-instruct': { input: 0.30, output: 0.30 },
+    'mixtral-8x7b-instruct': { input: 0.30, output: 0.30 },
+    'mixtral-8x22b-instruct': { input: 0.90, output: 0.90 }
+  }
+};
+
+// Helper function to get pricing for a model
+export function getStaticPricing(provider: string, modelId: string): { input: number; output: number } | null {
+  const providerPricing = staticPricing[provider];
+  if (!providerPricing) return null;
+  
+  // Check for exact match
+  if (providerPricing[modelId]) {
+    return providerPricing[modelId];
+  }
+  
+  // Check for default pricing
+  if (providerPricing['default']) {
+    return providerPricing['default'];
+  }
+  
+  return null;
+}

providers/together.ts

@@ -0,0 +1,59 @@
+import { BaseProviderFetcher } from './base';
+import { ProviderEntry, TogetherModel } from './types';
+
+export class TogetherFetcher extends BaseProviderFetcher {
+  name = 'together';
+
+  constructor(apiKey?: string) {
+    super('https://api.together.ai', apiKey, {
+      requestsPerMinute: 600  // Together rate limit from spec
+    });
+  }
+
+  async fetchModels(): Promise<ProviderEntry[]> {
+    try {
+      const response = await this.fetchWithRetry<TogetherModel[]>(
+        `${this.baseUrl}/v1/models`
+      );
+
+      return response.map(model => this.mapModelToProviderEntry(model));
+    } catch (error) {
+      console.error(`Failed to fetch Together models: ${error}`);
+      return [];
+    }
+  }
+
+  private mapModelToProviderEntry(model: TogetherModel): ProviderEntry {
+    const entry: ProviderEntry = {
+      provider: this.name,
+      context_length: model.context_length,
+      pricing: this.normalizePricing(
+        model.pricing.input,
+        model.pricing.output,
+        'per_million'
+      ),
+      owned_by: model.organization,
+      model_type: model.type
+    };
+
+    // Parse supported parameters from config if available
+    if (model.config) {
+      const configParams = this.parseConfigParameters(model.config);
+      Object.assign(entry, configParams);
+    }
+
+
+    return entry;
+  }
+
+  private parseConfigParameters(config: TogetherModel['config']): Partial<ProviderEntry> {
+    const result: Partial<ProviderEntry> = {};
+
+    // Check for stop sequences support
+    if (config.stop && config.stop.length > 0) {
+      result.supports_stop_sequences = true;
+    }
+
+    return result;
+  }
+}

providers/types.ts

@@ -0,0 +1,235 @@
+// Core provider data structure based on provider-api-spec.md
+export interface ProviderEntry {
+  provider: string;
+  status?: "live" | "offline" | "deprecated";
+  context_length?: number;
+  pricing?: {
+    input: number;   // $ per 1M tokens
+    output: number;  // $ per 1M tokens
+  };
+  quantization?: string;
+  max_completion_tokens?: number;
+  supported_parameters?: string[];
+  
+  // Model identification
+  model_id?: string;
+  created?: number;
+  
+  // Capability flags
+  supports_tools?: boolean;
+  supports_function_calling?: boolean;
+  supports_structured_output?: boolean;
+  supports_response_format?: boolean;
+  supports_streaming?: boolean;
+  supports_logprobs?: boolean;
+  supports_stop_sequences?: boolean;
+  supports_seed?: boolean;
+  supports_temperature?: boolean;
+  supports_top_p?: boolean;
+  supports_frequency_penalty?: boolean;
+  supports_presence_penalty?: boolean;
+  supports_repetition_penalty?: boolean;
+  supports_top_k?: boolean;
+  supports_min_p?: boolean;
+  supports_max_tokens?: boolean;
+  supports_logit_bias?: boolean;
+  supports_top_logprobs?: boolean;
+  supports_image_input?: boolean;
+  
+  // Performance metrics
+  latency_s?: number;
+  throughput_tps?: number;
+  performance_error?: string;
+  performance_tested_at?: string;
+  
+  // Additional metadata
+  owned_by?: string;
+  model_type?: string;
+  description?: string;
+  deprecated_at?: string;
+  model_class?: string;
+  is_gated?: boolean;
+}
+
+// Provider-specific response types
+export interface GroqModel {
+  id: string;
+  object: string;
+  created: number;
+  owned_by: string;
+  active: boolean;
+  context_window: number;
+  public_apps: any;
+  max_completion_tokens: number;
+}
+
+export interface CohereModel {
+  name: string;
+  endpoints: string[];
+  finetuned: boolean;
+  context_length: number;
+  tokenizer_url: string;
+  supports_vision: boolean;
+  features: string[];
+  default_endpoints: string[];
+  is_deprecated?: boolean;
+}
+
+export interface FireworksModel {
+  id: string;
+  object: string;
+  owned_by: string;
+  created: number;
+  kind: string;
+  supports_chat: boolean;
+  supports_image_input: boolean;
+  supports_tools: boolean;
+  context_length: number;
+}
+
+export interface FireworksDetailedModel {
+  name: string;
+  displayName: string;
+  description: string;
+  contextLength: number;
+  baseModelDetails: {
+    checkpointFormat: string;
+    defaultPrecision: string;
+    modelType: string;
+    moe: boolean;
+    parameterCount: string;
+    supportsFireattention: boolean;
+    tunable: boolean;
+    worldSize: number;
+  };
+  defaultSamplingParams: {
+    temperature?: number;
+    top_p?: number;
+    max_tokens?: number;
+    [key: string]: any;
+  };
+  supportsImageInput: boolean;
+  supportsLora: boolean;
+  supportsTools: boolean;
+  state: string;
+  deprecationDate: string | null;
+  huggingFaceUrl: string;
+  supportedPrecisions: string[];
+  deployedModelRefs: any[];
+}
+
+export interface TogetherModel {
+  id: string;
+  object: string;
+  created: number;
+  type: string;
+  display_name: string;
+  organization: string;
+  context_length: number;
+  pricing: {
+    input: number;    // $ per million tokens
+    output: number;   // $ per million tokens
+    hourly: number;
+    base: number;
+    finetune: number;
+  };
+  config: {
+    chat_template: string;
+    stop: string[];
+    bos_token: string;
+    eos_token: string;
+  };
+}
+
+export interface SambaNovaModel {
+  id: string;
+  object: string;
+  owned_by: string;
+  context_length: number;
+  max_completion_tokens: number;
+  pricing: {
+    prompt: string;      // $ per token
+    completion: string;  // $ per token
+  };
+  sn_metadata: any;
+}
+
+export interface NovitaModel {
+  id: string;
+  object: string;
+  created: number;
+  owned_by: string;
+  input_token_price_per_m: number;     // Cents per million tokens
+  output_token_price_per_m: number;    // Cents per million tokens
+  title: string;
+  description: string;
+  context_size: number;
+  max_output_tokens: number;
+  model_type: string;
+  features: string[];
+  endpoints: string[];
+  status: number;
+  display_name: string;
+}
+
+export interface FeatherlessModel {
+  id: string;
+  is_gated: boolean;
+  created: number;
+  model_class: string;
+  owned_by: string;
+  context_length: number;
+  max_completion_tokens: number;
+  available_on_current_plan: boolean;
+}
+
+// Minimal response types
+export interface CerebrasModel {
+  id: string;
+  object: string;
+  created: number;
+  owned_by: string;
+}
+
+export interface NebiusModel {
+  id: string;
+  created: number;
+  object: string;
+  owned_by: string;
+}
+
+export interface LambdaModel {
+  id: string;
+  object: string;
+  created: number;
+  owned_by: string;
+}
+
+// Base provider fetcher interface
+export interface ProviderFetcher {
+  name: string;
+  fetchModels(): Promise<ProviderEntry[]>;
+}
+
+// Configuration for rate limiting
+export interface RateLimitConfig {
+  requestsPerMinute?: number;
+  requestsPerHour?: number;
+  retryAttempts?: number;
+  initialBackoffMs?: number;
+}
+
+// Feature mapping types
+export interface FeatureMapping {
+  [key: string]: keyof ProviderEntry | string[] | null;
+}
+
+// Static pricing data structure
+export interface StaticPricing {
+  [provider: string]: {
+    [modelId: string]: {
+      input: number;   // $ per 1M tokens
+      output: number;  // $ per 1M tokens
+    };
+  };
+}

tsconfig.json

@@ -0,0 +1,27 @@
+{
+  "compilerOptions": {
+    // Enable latest features
+    "lib": ["ESNext", "DOM"],
+    "target": "ESNext",
+    "module": "ESNext",
+    "moduleDetection": "force",
+    "jsx": "react-jsx",
+    "allowJs": true,
+
+    // Bundler mode
+    "moduleResolution": "bundler",
+    "allowImportingTsExtensions": true,
+    "verbatimModuleSyntax": true,
+    "noEmit": true,
+
+    // Best practices
+    "strict": true,
+    "skipLibCheck": true,
+    "noFallthroughCasesInSwitch": true,
+
+    // Some stricter flags (disabled by default)
+    "noUnusedLocals": false,
+    "noUnusedParameters": false,
+    "noPropertyAccessFromIndexSignature": false
+  }
+}