AI Models Directory

Excellent for coding, reasoning, and multilingual tasks.

Meta's latest. Great balance of speed and capability.

Best general-purpose chat model with strong reasoning.

Specialized reasoning model with chain-of-thought capabilities.

Efficient reasoning model, great for laptops.

The OG local model. Fast, capable, battle-tested.

Mixture of Experts. GPT-3.5 level performance locally.

Google's best open model. Excellent instruction following.

Meta's dedicated coding model. Great for completions.

Popular vision model. Good balance of speed and quality.

Code specialist with excellent completion and generation.

Meta's dedicated coding model. Great for completions.

Top-tier coding model. Rivals GPT-4 for code.

Qwen fine-tuned for code. Excellent all-rounder.

Open-source coding model trained on The Stack v2.

Image understanding and visual Q&A capabilities.

Tiny, fast vision model. Great for quick image tasks.

Popular vision model. Good balance of speed and quality.

Great for laptops with limited RAM.

Ultra-small. Runs on anything.

Smallest Qwen. Perfect for edge devices.

Meta's latest. Great balance of speed and capability.

The largest open model ever. GPT-4 class performance.

685B MoE beast. Rivals Claude and GPT-4.

176B total params. The bigger Mixtral brother.

UAE sovereign model. 180B dense transformer.

104B params. Enterprise-grade RAG specialist.

132B MoE from Databricks. Enterprise beast.

Elon's 314B MoE. The Twitter/X AI.