mistral-medium-2508: benchmark scores, price and ranking
mistral-medium-2508 from Mistral AI ranks #666 of 743 LLMs on the LLMs Tiger aggregate benchmark score (43.7), based on 6 benchmark results captured up to 2026-10-13.
- Organisation
- Mistral AI
- Type
- Frontier model
- Open weights
- No
- Released
- 2025-08-12 (source: epoch)
- Aggregate rank
- #666 of 743 (score 43.7)
Relative to other models, mistral-medium-2508 is strongest on DTBench (#79 of 146) and weakest on ALE-Bench (#77 of 79).
Across the 6 benchmarks where at least ten models are scored, mistral-medium-2508 is in the top 10% on 0 and in the top half on 0.
mistral-medium-2508 benchmark results
| Benchmark | Score | Rank | Source | Captured |
|---|---|---|---|---|
| DTBench | 67.5% | #79 of 146 | Epoch AI Benchmarking Hub | 2026-10-13 |
| LMCA | 19.1% | #74 of 108 | Epoch AI Benchmarking Hub | 2026-10-13 |
| WeirdML | 33.1% | #84 of 109 | Epoch AI Benchmarking Hub | 2026-10-13 |
| SciCode | 33.8% | #86 of 106 | Epoch AI Benchmarking Hub | 2026-10-13 |
| CritPt | 0.0% | #108 of 115 | Epoch AI Benchmarking Hub | 2026-10-13 |
| ALE-Bench | 210 | #77 of 79 | Epoch AI Benchmarking Hub | 2026-10-13 |
About mistral-medium-2508
mistral-medium-2508 is a frontier model from Mistral AI. It was released on 2025-08-12, according to Epoch AI. No API price is listed for it yet. We hold 6 benchmark results for it across 4 areas, and none is in the top tenth of its board.
Where mistral-medium-2508 stands
On our aggregate score, mistral-medium-2508 is #666 of 743 ranked models, in the lower half. Its score is 43.7; the scale is centred on 50, so it sits 6.3 points below the typical model. That beats 10% of the ranked field. Among the 44 ranked Mistral AI models it is #33. The next one up is Magistral Small 1.2 at 43.8. The next one down is Mixtral-8x22B-Instruct-v0.1 at 43.5.
mistral-medium-2508, benchmark by benchmark
mistral-medium-2508 lands below the middle on DTBench: 67.5%, #79 of 146, with 53% of the field scoring higher. Claude Opus 5.5 (max) leads at 98.9%; the closest result above is Claude 3.5 Sonnet (Jun 2024) at 67.8%. It covers dynamic tool-use benchmark. The board median is 68.8%, so mistral-medium-2508 is 1.3 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
mistral-medium-2508 lands below the middle on LMCA: 19.1%, #74 of 108, with 68% of the field scoring higher. Claude Opus 5.5 (max) leads at 68.2%; the closest result above is qwen3-30b-a3b-thinking-2507 at 19.5%. It covers language-model capability assessment. The board median is 29.2%, so mistral-medium-2508 is 10.2 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
mistral-medium-2508 lands below the middle on WeirdML: 33.1%, #84 of 109, with 76% of the field scoring higher. GPT-6 Astra (max) leads at 93.3%; the closest result above is qwen3-coder-next at 34.4%. It covers unusual machine-learning tasks. The board median is 41.2%, so mistral-medium-2508 is 8.0 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
SciCode is a weak spot. mistral-medium-2508 scores 33.8%, #86 of 106, behind 80% of the models measured. The leader, Claude Opus 5.5 (max), reaches 66.9%. It covers scientific research code. The board median is 44.3%, so mistral-medium-2508 is 10.5 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
CritPt is a weak spot. mistral-medium-2508 scores 0.0%, #108 of 115, behind 93% of the models measured. The leader, GPT-5.6 Sol (max), reaches 32.3%. It covers unpublished research-level physics problems. The board median is 2.9%, so mistral-medium-2508 is 2.9 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
ALE-Bench is a weak spot. mistral-medium-2508 scores 210, #77 of 79, behind 96% of the models measured. The leader, GPT-6 Astra (max), reaches 2,951. It covers AtCoder heuristic-optimisation contests (performance rating). The board median is 764, so mistral-medium-2508 is 554 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Strengths and weaknesses
Software engineering: weak, ahead of 11% of the field on average across 2 benchmarks. Its best board here is SciCode (#86 of 106) and its weakest is ALE-Bench (#77 of 79). General knowledge and chat: weak, ahead of 27% of the field on average across 2 benchmarks. Its best board here is LMCA (#74 of 108) and its weakest is WeirdML (#84 of 109).
Release history
The Mistral AI model released before it was codestral-2508 on 2025-07-30, which scores 46.8. The next one was Magistral Medium 1.2 on 2025-09-18, at 47.1.
Reading these numbers
The 6 scores on this page come from 1 source: Epoch AI Benchmarking Hub (6). No official leaderboard supplied any of them, so treat them as third-party measurements. With 6 results the aggregate rank can move a lot when one new score arrives.
Models ranked nearby
- #663 Magistral Small 1.2 (43.8)
- #664 GPT-4 (Mar 2023) (43.7)
- #665 gemma-2-27b (43.7)
- #667 Claude 2 (43.7)
- #668 Llama-3.1-405B-Instruct (43.6)
- #669 granite-3.1-2b-base (43.6)
More from Mistral AI
- Mistral Large 2 (2407) (#231, 49.9)
- codestral-2405 (#352, 48.4)
- pixtral-large-2411 (#360, 48.3)
- Codestral-22B-v0.1-Q4_K_M (#392, 48.0)
- codestral:22b-v0.1-q8_0 (#393, 48.0)
- Mistral Large 4 (high) (#411, 47.7)
Compare it on the live leaderboard → · How the aggregate score works