Mistral Medium 3 (25.05)
Version: 1
Models from Microsoft, Partners, and Community
Models from Microsoft, Partners, and Community models are a select portfolio of curated models both general-purpose and niche models across diverse scenarios by developed by Microsoft teams, partners, and community contributors- Managed by Microsoft: Purchase and manage models directly through Azure with a single license, world class support and enterprise grade Azure infrastructure
- Validated by providers: Each model is validated and maintained by its respective provider, with Azure offering integration and deployment guidance.
- Innovation and agility: Combines Microsoft research models with rapid, community-driven advancements.
- Seamless Azure integration: Standard Azure AI Foundry experience, with support managed by the model provider.
- Flexible deployment: Deployable as Managed Compute or Serverless API, based on provider preference.
Key capabilities
About this model
Mistral Medium 3 is a SOTA & versatile model designed for a wide range of tasks, including programming, mathematical reasoning, understanding long documents, summarization, and dialogue.Key model capabilities
- Programming
- Math reasoning
- Dialogue
- Long document understanding
- Visual understanding
- Summarization
- Low-latency applications
- Multi-modal capabilities
- Function calling
Use cases
See Responsible AI for additional considerations for responsible use.Key use cases
Mistral Medium 3 (25.05) is a great versatile model for tasks such as:- Programming
- Math reasoning
- Dialogue
- Long document understanding
- Visual understanding
- Summarization
- Low-latency applications
Out of scope use cases
The provider has not supplied this information.Pricing
Pricing is based on a number of factors, including deployment type and tokens used. See pricing details here.Technical specs
The provider has not supplied this information.Training cut-off date
The provider has not supplied this information.Training time
The provider has not supplied this information.Input formats
It boasts multi-modal capabilities, enabling it to process visual inputs, and supports dozens of languages, including over 80 coding languages.Output formats
The provider has not supplied this information.Supported languages
supports dozens of languages, including over 80 coding languagesSample JSON response
The provider has not supplied this information.Model architecture
The provider has not supplied this information.Long context
Mistral Medium 3 is optimized for single-node inference, particularly for long-context applications. Its size allows it to achieve high throughput on a single node.Optimizing model performance
The provider has not supplied this information.Additional assets
The provider has not supplied this information.Training disclosure
Training, testing and validation
The provider has not supplied this information.Distribution
Distribution channels
The provider has not supplied this information.More information
The provider has not supplied this information.Responsible AI considerations
Safety techniques
The provider has not supplied this information.Safety evaluations
The provider has not supplied this information.Known limitations
The provider has not supplied this information.Acceptable use
Acceptable use policy
The provider has not supplied this information.Quality and performance evaluations
Source: Mistral AIAcademic Evals
Coding
| Benchmark | Mistral Medium 3 | Llama 4 Maverick | GPT-4o | Claude Sonnet 3.7 | Command-A | DeepSeek 3.1 |
|---|---|---|---|---|---|---|
| HumanEval 0-shot | 0.921 | 0.854 | 0.915 | 0.921 | 0.829 | 0.933 |
| LiveCodeBench (v6) 0-shot | 0.303 | 0.287 | 0.314 | 0.360 | 0.263 | 0.429 |
| MultiPL-E average 0-shot | 0.814 | 0.764 | 0.798 | 0.834 | 0.731 | 0.849 |
Instruction Following
| Benchmark | Mistral Medium 3 | Llama 4 Maverick | GPT-4o | Claude Sonnet 3.7 | Command-A | DeepSeek 3.1 |
|---|---|---|---|---|---|---|
| ArenaHard 0-shot | 0.971 | 0.918 | 0.954 | 0.932 | 0.951 | 0.973 |
| IfEval 0-shot | 0.894 | 0.889 | 0.872 | 0.918 | 0.897 | 0.891 |
Math
| Benchmark | Mistral Medium 3 | Llama 4 Maverick | GPT-4o | Claude Sonnet 3.7 | Command-A | DeepSeek 3.1 |
|---|---|---|---|---|---|---|
| Math500 Instruct 0-shot | 0.910 | 0.900 | 0.764 | 0.830 | 0.820 | 0.938 |
Knowledge
| Benchmark | Mistral Medium 3 | Llama 4 Maverick | GPT-4o | Claude Sonnet 3.7 | Command-A | DeepSeek 3.1 |
|---|---|---|---|---|---|---|
| GPQA Diamond 5-shot CoT | 0.571 | 0.611 | 0.525 | 0.697 | 0.465 | 0.611 |
| MMLU Pro 5-shot CoT | 0.772 | 0.804 | 0.758 | 0.800 | 0.689 | 0.811 |
Long Context
| Benchmark | Mistral Medium 3 | Llama 4 Maverick | GPT-4o | Claude Sonnet 3.7 | Command-A | DeepSeek 3.1 |
|---|---|---|---|---|---|---|
| RULER 32K | 0.960 | 0.948 | 0.960 | 0.957 | 0.956 | 0.958 |
| RULER 128K | 0.902 | 0.867 | 0.889 | 0.938 | 0.912 | 0.919 |
Multimodal
| Benchmark | Mistral Medium 3 | Llama 4 Maverick | GPT-4o | Claude Sonnet 3.7 | Command-A | DeepSeek 3.1 |
|---|---|---|---|---|---|---|
| MMMU 0-shot | 0.661 | 0.718 | 0.661 | 0.713 | - | - |
| DocVQA 0-shot | 0.953 | 0.941 | 0.859 | 0.843 | - | - |
| AI2D 0-shot | 0.937 | 0.844 | 0.933 | 0.788 | - | - |
| ChartQA 0-shot | 0.826 | 0.904 | 0.860 | 0.763 | - | - |
Human Evals
Mistral Wins vs Llama 4 Maverick
| Domain | Mistral Win Rate | Llama 4 Maverick Win Rate |
|---|---|---|
| Coding | 81.82 | 18.18 |
| Multimodal | 53.85 | 46.15 |
| English | 66.67 | 33.33 |
| French | 71.43 | 28.57 |
| Spanish | 73.33 | 26.67 |
| German | 62.50 | 37.50 |
| Arabic | 64.71 | 35.29 |
Mistral Wins vs Competitor Wins for Coding
| Model | Mistral Wins | Other Model Wins |
|---|---|---|
| claude_3_7 | 40.00 | 60.00 |
| deepseek_v3_1 | 37.50 | 62.50 |
| gpt_4o | 50.00 | 50.00 |
| command_a | 69.23 | 30.77 |
| llama_4_maverick | 81.82 | 18.18 |
Benchmarking methodology
Source: Mistral AI The provider has not supplied this information.Public data summary
Source: Mistral AI The provider has not supplied this information.Model Specifications
Context Length128000
Quality Index0.77
LicenseCustom
Last UpdatedOctober 2025
Input TypeText,Image
Output TypeText
ProviderMistral AI
Languages27 Languages
Related Models
