aquif-moe-400m

aquif-moe-400m is our compact Mixture of Experts (MoE) model, with only 400 million active parameters. It offers impressive performance-per-VRAM efficiency, making it a strong choice for resource-limited setups.

Model Overview

• Name: aquif-moe-400m
• Parameters: 400 million active parameters (1.3 billion total)
• Context Window: 128,000 tokens
• Architecture: Mixture of Experts (MoE)
• Type: General-purpose LLM
• Hosted on: Ollama, Huggingface

Key Features

• Highly efficient VRAM utilization (77.0 performance points per GB)
• Expansive 128K token context window for handling long documents
• Competitive performance despite fewer parameters
• Optimized for local inference on consumer hardware
• Ideal for resource-constrained environments
• Supports high-throughput concurrent sessions

Performance Benchmarks

aquif-moe-400m delivers solid performance across multiple benchmarks, especially for its size:

Benchmark	aquif-moe (0.4b)	Qwen 2.5 (0.5b)	Gemma 3 (1b)
MMLU	26.6	45.4	26.5
HumanEval	32.3	22.0	8.1
GSM8K	33.9	36.0	6.1
Average	30.9	34.4	11.3

VRAM Efficiency

aquif-moe-400m excels in VRAM efficiency:

Model	Average Performance	VRAM (GB)	Performance per VRAM
aquif-moe	30.9	0.4	77.3
Qwen 2.5	34.4	0.6	57.3
Gemma 3	11.3	1.0	11.3

Use Cases

• Edge computing and resource-constrained environments
• Mobile and embedded applications
• Local development environments
• Quick prototyping and testing
• Personal assistants on consumer hardware
• Enterprise deployment with multiple concurrent sessions
• Long document analysis and summarization
• High-throughput production environments

Limitations

• No thinking mode capability
• May show hallucinations in some areas
• May struggle with more complex reasoning tasks
• Not optimized for specialized domains

Getting Started

To run via Ollama:

ollama run aquiffoo/aquif-moe-400m
``´