Zen Training
Train Zen models with HuggingFace datasets
None defined yet.
Democratize AI while protecting our planet
A groundbreaking collaboration between Hanzo AI (Techstars-backed, award-winning GenAI lab) and Zoo Labs Foundation (501(c)(3) environmental non-profit), eco-friendly AI that runs entirely on your device ā no cloud, no subscriptions, no surveillance.
| Model | Size | Description |
|---|---|---|
| zen-nano | 0.6B | Ultra-lightweight edge model |
| zen-eco-instruct | 4B | Efficient instruction-following |
| zen-eco-thinking | 4B | Chain-of-thought reasoning |
| zen-agent | 4B | Tool-calling and functions |
| zen-coder | 480B | Code generation and analysis |
| zen-next | 80B | Professional-grade LLM |
| zen-omni | 30B Multimodal | Vision-language understanding |
| Model | Type | Description |
|---|---|---|
| zen-3d | 3.3B | Controllable 3D asset generation |
| zen-voyager | Diffusion | Camera-controlled world exploration |
| zen-world | Diffusion | Large-scale world simulation |
| Model | Type | Description |
|---|---|---|
| zen-director | 5B | Text/image-to-video generation |
| zen-i2v | Diffusion | Image-to-video animation |
| zen-video | Diffusion | High-quality video synthesis |
| Model | Type | Description |
|---|---|---|
| zen-musician | 7B | Music generation (lyrics ā songs) |
| zen-foley | Diffusion | Video-to-audio Foley effects |
| Model | Type | Description |
|---|---|---|
| zen-scribe | 4B | Audio transcription |
| Model | Type | Description |
|---|---|---|
| zen-artist | Multimodal | Creative visual generation |
| zen-designer-instruct | Multimodal | Design instruction following |
| zen-designer-thinking | Multimodal | Design reasoning |
| Model | Type | Description |
|---|---|---|
| zen-agent | 4B | Function calling and tool using agent |
| zen-coder-4B | 4B | Advanced coding assistant |
| zen-coder-80B | 80B | Documentation generation |
| zen-coder-480B | 480B | Code action suggestions |
| Model | Type | Description |
|---|---|---|
| zen-guard | Classification | Content safety detection |
| zen-guard-gen | 4B | Safe content generation |
| zen-guard-stream | Classification | Real-time |
| safety monitoring |
| Model | Type | Description |
|---|---|---|
| zen-embedding | Embedding | High-quality text embeddings |
š Ultra-Efficient - 06.B parameters to 480B-class performance ⢠Runs on phones, laptops, and AI super computers š Truly Private - 100% local processing ⢠No accounts, no telemetry, no tracking š± Environmentally Responsible - 95% less energy than cloud AI ⢠Carbon-negative operations š Free Forever - Apache 2.0 licensed ⢠No premium tiers or API fees
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("zenlm/zen-eco-instruct")
tokenizer = AutoTokenizer.from_pretrained("zenlm/zen-eco-instruct")
inputs = tokenizer("Explain quantum computing", return_tensors="pt")
outputs = model.generate(**inputs, max_length=200)
print(tokenizer.decode(outputs[0]))
---
šļø Infrastructure
https://github.com/zenlm/zen-gym - Unified training platform supporting LoRA, GRPO, DPO, and all
modern methods
https://github.com/zenlm/zen-engine - High-performance inference engine (44K tok/s) with
OpenAI-compatible API
---
š License
Models: Apache 2.0 ⢠Code: MIT License ⢠Privacy: No data collection, ever
---
šļø Organizations
Hanzo AI Inc - Techstars Portfolio ⢠Award-winning GenAI lab ⢠https://hanzo.ai
Zoo Labs Foundation - 501(c)(3) Non-Profit ⢠Environmental preservation ⢠https://zoolabs.io
---
š® Contact
š https://zenlm.org ⢠š¬ https://discord.gg/hanzoai ⢠š¦ https://twitter.com/hanzoai ⢠š§
[email protected]