Mistral AI model product
Trained solely with reinforcement learning on top of Mistral Medium 3, Magistral Medium is a reasoning model that achieves strong performance on complex math and code tasks without relying on distillation from existing reasoning models.
Updated Aug 12, 2026. Default version: Magistral Medium
Technical details for the model's default version.
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Benchmark scores for Magistral Medium.
| Aider-Polyglot | 47.1% | 18 | 22 | 19.1% | C | |
| AIME 2024 | 73.6% | 36 | 53 | 32.7% | C | |
| LiveCodeBench | 50.3% | 44 | 73 | 40.3% | C | |
| Humanity's Last Exam | 9.0% | 83 | 93 | 10.9% | C | |
| AIME 2025 | 64.9% | 93 | 114 | 18.6% | C | |
| GPQA | 70.8% | 126 | 234 | 46.4% | C |
Preference and agent-evaluation results for the default version.
| text style control | overall | 255 | 1304.0 | 11,677 | N/A | |
| text | overall | 264 | 1254.4 | 11,677 | N/A |
Provider-specific output speed and catalog latency for Magistral Medium. Runtime does not affect the capability score.
No provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Official vendor API pricing appears first, followed by individual provider offers.
The default version has no current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Available versions of this model. The score column identifies the version used in the overall ranking.
| Magistral Medium | 19.8 | 24B | N/A | N/A | No | Apache 2.0 |
Key information about Magistral Medium and its available data.
Trained solely with reinforcement learning on top of Mistral Medium 3, Magistral Medium is a reasoning model that achieves strong performance on complex math and code tasks without relying on distillation from existing reasoning models.
The training uses an RLVR framework with modifications to GRPO, enabling improved reasoning ability and multilingual consistency.
Data as of 2026-08-11.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
Common questions about Magistral Medium.
Magistral Medium's default version was released on Jun 10, 2025.
No official standard PAYG price is currently available for Magistral Medium.
Magistral Medium was created by Mistral AI.
A context window is not available for the default version.
No. The default version is not marked as having publicly available weights.
No provider offering is currently linked to the default version.
Nearby ranked alternatives include Min istral 3, Gemini 1.5 Pro, Magistral Small.