BenchLeader
Menu
LeaderboardsModelsProvidersBenchmarksCompareReleases3BlogMethodology
Mistral AIAuto-detected

Mistral Small 3.1

Mistral Small 3.1 is a Mistral AI proprietary model, released 17 Mar 2025. Its best configuration ranks #616 of 372 on the BenchLeader Index at 34.6 ±4.3, in the lower half. It scores highest in reasoning (37) and lowest in coding (22). At $0.150 per million tokens blended it is among the cheapest fifth of ranked models. Last measured 10 Sept 2026.

Blended price
$0.150/M
$0.100 in · $0.300 out
Output speed
–
First answer
–
Context
128k
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index35
  2. Reasoning37
  3. Coding22
  4. Maths34
  5. Knowledge31

Versions

Mistral AI has shipped 8 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

ModelReleasedIndexRank
Mistral Small 416 Mar 202647.2#364
Mistral Small 3.117 Mar 202539.8#535
Mistral Small 3.1this page17 Mar 202534.6#616
Mistral Small 330 Jan 202538.4#565
Mistral Small 325 Jan 2025––
Mistral Small 3.2–42.3#474
Mistral Small 2402–32.0#623
Mistral Small 3–––

Benchmark results

One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

BenchmarkdefaultSource
GPQA Diamond47.5%#213Epoch AI Benchmarking Hub
GPQA Diamond (Vals)not in index44.2%#123Vals AI

Coding

BenchmarkdefaultSource
SciCode26.5%#149SciCode
LiveCodeBench31.8%#133Vals AI

Maths

BenchmarkdefaultSource
OTIS Mock AIME5.8%#229Epoch AI Benchmarking Hub
MATH Level 546.8%#55Epoch AI Benchmarking Hub
AIME (Vals)3.5%#89Vals AI
MGSM84.2%#66Vals AI

Knowledge

BenchmarkdefaultSource
MMLU-Pro66.0%#122Vals AI
LegalBench69.2%#118Vals AI
CorpFin44.2%#112Vals AI
TaxEval58.3%#129Vals AI
MedQA69.1%#78Vals AI

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0001––
Summarise a 30-page report12,000 / 600$0.0014––
Code edit6,000 / 1,500$0.0011––
Agentic coding session60,000 / 4,000$0.0072––
Structured extraction2,000 / 200$0.0003––

See also

Best under $2/MBest under $0.50/MBest for codingAll Mistral AI models

Data as of 10 Sept 2026. Compare with another model.