Every model worth running, side by side.

Google · gemma

Gemma 4 26B A4B IT

Google DeepMind's instruction-tuned Mixture-of-Experts model with 25.2B total parameters and only 3.8B active per token, delivering high-tier reasoning and vision understanding across a 256K-token context window — available as open weights under Apache 2.0.

Google · gemma

Gemma 4 31B IT

Google DeepMind's open-weights 30.7B dense multimodal model with a 256K-token context window, optional chain-of-thought thinking, and native function calling — instruction-tuned for complex, long-context, and multilingual tasks.

Mistral Ai · mistral-small

mistral-small-2603

Mistral AI's 119B open-weights MoE model unifying instruct, on-demand reasoning, vision, and coding in a single Apache 2.0 checkpoint, with a 262K-token context window and interactive-class latency.