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.

€0.09/€0.28/M in·out

Mistral Ai · mistral-large

mistral-large-2512

Mistral AI's December 2025 open-weight flagship — a sparse MoE with 675B total / 41B active parameters, 256k context, native vision, and Apache 2.0 licence — built for complex instruction-following, agentic workflows, and multilingual deployments.

€1.84/€5.52/M in·out

Mistral Ai · mistral-nemo

open-mistral-nemo

A 12B open-weights instruction model co-developed by Mistral AI and NVIDIA, featuring a 131k-token context window, the efficient Tekken tokenizer, native tool calling, and multilingual support across nine languages — available under Apache 2.0 for unrestricted self-hosted inference.

€0.13/€0.13/M in·out

No models match your search.

Model of the month

mistral-small-2603 — Mistral AI's 119B open-weights MoE model unifying instruct, on-demand reasoning, vision, and coding in a single Apache 2.

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.

Stav Hosted119B (6.5B active) parameters262K contextOpen weights

Read the review See endpoints →

mistral-mistral-small-2603

EU-Hosted · Mistral

€0.18/M blended

The router

One endpoint. The whole catalogue. Sovereignty-first by default.

Smart Router classifies each prompt and picks the right model under your sovereignty preference — sub-15ms overhead.

Read the spec→