Magistral models

Strongest: Magistral Medium 1.2

Mistral deprecated Magistral Medium 1.2 on 2026-05-22 with a retirement date of 2026-07-31 and points reasoning users to Mistral Medium 3.5, which has reasoning built in; the model is gone from Mistral's API pricing page, and as a closed-weight model it has no self-host path. Magistral Medium 1.2 (alias magistral-medium-2509, shipped 18 September 2025) was Mistral's flagship reasoning model: a Premier/closed-weight, multimodal model that produces visible chain-of-thought before answering, with a 40,960-token output ceiling for long traces. It posts AIME24 91.82% / AIME25 83.48% and GPQA Diamond 76.26%, and its reasoning traces stay in the user's language — a genuine multilingual-reasoning differentiator. It was priced at $2.00/$5.00. The buyer's sentence: it was the right pick for auditable, multilingual hard-reasoning routing until Medium 3.5's reasoning knob absorbed the niche and replaced it.