Mistral AI
4 ranked models in a catalog of 32 text models. Mistral Large 4 leads this lab at #53 overall.
Ranked models
Global ranks · current snapshot| Rank | Model | SI Score | coding | math | reasoning | preference | Confidence | Blended price | Context | Released |
|---|---|---|---|---|---|---|---|---|---|---|
| #53 | Mistral Large 4 | 62.2 | 65.2 | 88.1 | 77.4 | 75.7 | 64% | $1.03 | 1M | Oct 6, 2026 |
| #111 | Mistral Small 3.1 24BOpen weights | 48.4 | — | 33.1 | 47.5 | 59.5 | 64% | — | 128K | Mar 17, 2025 |
| #135 | Mistral Large 2.1Open weights | 35.3 | 6.1 | 36.1 | 51.3 | 58.1 | 100% | — | 131K | Nov 18, 2024 |
| #136 | Mistral Medium 3 | 33.0 | 3.8 | 65.2 | 22.9 | 45.7 | 85% | — | 131K | May 7, 2025 |
Pillars use a 0–100 scale. * marks an imputed neutral prior where the model has no scored results in that pillar; it is not a measured benchmark result. Prices are USD per 1M tokens; see the blend and price comparison.
Scores and releases
Provisional models
28 awaiting rank-eligible evidence- Mistral Medium (latest)
Results cover 1 pillar; at least 2 required
- Mistral Medium 3.5
Evidence completeness 37.0%; at least 50% required
- Mistral Small (latest)
No usable open benchmark results found
- Mistral Small 4
Results cover 1 pillar; at least 2 required
- Devstral 2
Results cover 1 pillar; at least 2 required
- Devstral Small 2
No usable open benchmark results found
- Devstral 2 (latest)
No usable open benchmark results found
- Ministral 14B
No usable open benchmark results found
- Ministral 3 14B
No usable open benchmark results found
- Ministral 3 3B
No usable open benchmark results found
- Ministral 3 8B
No usable open benchmark results found
- Mistral Large 3
Evidence completeness 37.0%; at least 50% required
- Magistral Small 1.2
Evidence completeness 32.0%; at least 50% required
- Codestral 25.08
No usable open benchmark results found
- Devstral Medium
Results cover 1 pillar; at least 2 required
- Devstral Small
Results cover 1 pillar; at least 2 required
- Mistral Small 3.2
Results cover 1 pillar; at least 2 required
- Magistral Small
Evidence completeness 48.0%; at least 50% required
- Pixtral Large (25.02)
No usable open benchmark results found
- Magistral Medium (latest)
Results cover 1 pillar; at least 2 required
- Mistral Large (latest)
No usable open benchmark results found
- Pixtral Large (latest)
No usable open benchmark results found
- Ministral 3B
No usable open benchmark results found
- Ministral 8B Instruct
No usable open benchmark results found
- Pixtral 12B
No usable open benchmark results found
- Mistral Nemo
No usable open benchmark results found
- Codestral (latest)
No usable open benchmark results found
- Codestral-22B-v0.1
No usable open benchmark results found
Latest Mistral AI news
All AI news ↗Introducing Mistral Large 4
Mistral Large 4 now ranks #53
Hallo, Deutschland!
Mistral and Mozilla are bringing open, private and multilingual AI to your web browser
Cloudera and Mistral Partner to Bring Specialized, Sovereign Intelligence to Enterprise Data
Modernizing complex legacy code with AI agents.