Skip to main content
Models/Mistral AI/Magistral Medium

Magistral Medium

Mistral AI 🧠 Language Model

Advanced reasoning model for complex multi-step logic and problem-solving.

⏹ Retired

Specifications

Input modalitiestext
Output modalitiestext
Capabilitieschat, reasoning
Relative speed●●●○○ Average
Deprecated on2026-08-09
Context window131,072 tokens
Max output40,000 tokens
Vision inputNo
Tool callingNo
Streaming outputYes
Input price$3.50 / 1M
Output price$8.75 / 1M

Pricing

$3.50 / 1M input tokens  ·  $8.75 / 1M output tokens

Platform credit pricing — live rates on the Models page.

More from Mistral AI

M
Mistral Large 3
Open-weight flagship. 675B MoE (41B active), multimodal with vision. Apache 2.0 licensed.
🧠 Language Model
M
Mistral Large 4
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.
🧠 Language Model
M
Mistral Small
Fast and efficient model for simple tasks. Great balance of performance and cost.
🧠 Language Model
C
Codestral
Specialized model for code generation, completion, and understanding.
🧠 Language Model