Models

Small, fast Meta model good for on-device chat and lightweight tasks.

Popular 8B general-purpose model; strong quality-to-size balance.

Strong multilingual 7B model with a permissive license.

Code-specialized 14B model competitive on programming benchmarks.

Google's efficient 9B open model, strong for its size.

Microsoft's compact 3.8B model tuned for reasoning in a small footprint.

Efficient, permissively-licensed 7B model widely used as a local baseline.

Distilled reasoning model that exposes step-by-step thinking, runs locally.

Large 70B model with near-frontier quality; needs a big GPU or lots of RAM.