Mistral Large 4
Mistral AI
🧠 Language Model
Mistral's open-weight flagship (public preview, 2026-10-06). 1.05T MoE (52B active) with a vision encoder; reasoning on demand, tools. 1M context.
Specifications
Input modalitiestext, image
Output modalitiestext
Capabilitieschat, tools, reasoning
Relative speed●●●○○ Average
Context window1,048,576 tokens
Max output262,144 tokens
Vision inputYes
Tool callingYes
Input price$1.19 / 1M
Output price$3.66 / 1M
Pricing
$1.19 / 1M input tokens · $3.66 / 1M output tokens
Platform credit pricing — live rates on the Models page.