Mistral Small 3

January 30, 2025Last synced Jul 10, 2026, 8:59 PM

Human Frontier

—

Human-calibrated frontier signal, backed by Arena-style preference evidence and separate from raw AgMoBench benchmark composite scores.

Blended Price

$0.15/M

Input Price

$0.10/M

Output Price

$0.30/M

Speed

152 tok/s

TTFT

0.52s

Benchmark Scores

Intelligence Index

Performance Radar

Mistral Small 3

Field Average

External Benchmarks

AA-Omniscience Accuracy

How Mistral Small 3 Compares

Axes

X Axis

Y Axis

Bubble Size

Filters

Blended Price (USD)$0.00 – $37.5

AgMoBench Overall≥ 1.9

Providers

ai2ai21-labsalibabaanthropicarceeawsazurebaidu

Show quadrants

google

mistral

deepseek

stepfun

openai

anthropic

liquidai

meta

cohere

nvidia

azure

kimi

minimax

alibaba

ibm

nous-research

aws

multiversecomputing

zai

deepcogito

kwaikat

xai

xiaomi

ai21-labs

inclusionai

baidu

sarvam

bytedance_seed

upstage

reka-ai

inception

arcee

tencent

ai2

perplexity

swiss-ai-initiative

Bubble size = Context Window

8191.00

2000000.00

Compare with other models

GPQA Diamond

0.5 / 100

LiveCodeBench

0.3 / 100

Terminal-Bench Hard

— / 100

τ²-Bench

0.2 / 100

AA-Omniscience Hallucination RatePredicted

95.1 / 100

Aider PolyglotPredicted

20.3 / 100

AlpacaEval 2.0 LCPredicted

32.6 / 100

AlpacaEval 2.0 RawPredicted

25.6 / 100

ARC-AGI-1Predicted

6.4 / 100

ARC-AGI-1 Cost per TaskPredicted

0.0

ARC-AGI-2Predicted

0.0 / 100

ARC-AGI-2 Cost per TaskPredicted

0.0

BFCL (Berkeley Function Calling)Predicted

32.1

BigCodeBench CompletePredicted

49.0 / 100

BigCodeBench InstructPredicted

39.6 / 100

AA Intelligence Index (Matrix)Predicted

63.0

AA Long Context Reasoning (Matrix)Predicted

71.6

AIME 2024Predicted

39.3

AIME 2025 (Matrix)Predicted

42.9

Arena-Hard AutoPredicted

46.3

BrowseCompPredicted

66.0

BRUMO 2025Predicted

88.3

CMIMC 2025Predicted

75.4

GPQA Diamond (Matrix)Predicted

53.8

HLE (Matrix)Predicted

14.4

HMMT Feb 2025Predicted

43.3

HMMT Nov 2025Predicted

89.9

HumanEvalPredicted

80.3

IFBench (Matrix)Predicted

32.2

IFEvalPredicted

79.5

IMO 2025Predicted

14.8

LiveCodeBench (Matrix)Predicted

42.7

MATH-500 (Matrix)Predicted

90.8

MathArena Apex 2025Predicted

1.3

MMLU-Pro (Matrix)Predicted

67.2

MMMU-ProPredicted

79.3

MRCR v2Predicted

75.7

OSWorldPredicted

34.6

SimpleQAPredicted

24.5

SMT 2025Predicted

81.0

SWE-bench ProPredicted

27.9

Tau-Bench Telecom (Matrix)Predicted

96.3

Terminal-Bench 2.0Predicted

26.5

Terminal-Bench 1.0Predicted

15.8

USAMO 2025Predicted

6.0

Video-MMUPredicted

85.7

browsecompPredicted

64.6

BullshitBenchPredicted

16.0 / 100

Aider PolyglotPredicted

0.0

Apex AgentsPredicted

2.5

Arc Agi 2Predicted

0.0

BIG-Bench HardPredicted

3.0

CAD-EvalPredicted

6.9

Chess PuzzlesPredicted

0.1

CyBenchPredicted

0.1

DeepResearchBenchPredicted

0.3

FictionLiveBenchPredicted

0.3

GeoBenchPredicted

0.0

GSM8K (Epoch)Predicted

5.7

HellaSwagPredicted

5.2

Lech Mazur WritingPredicted

7.0

METR Time HorizonsPredicted

0.6

OTIS Mock AIME 2024–2025Predicted

0.1

PosttrainbenchPredicted

0.0

SimpleQA Verified (Epoch)Predicted

0.3

The Agent CompanyPredicted

1.2

TriviaQAPredicted

3.5

WinoGrandePredicted

0.8

FrontierMathPredicted

15.2 / 100

GAIA Level 1Predicted

23.9

GAIA Level 2Predicted

6.5

GAIA Level 3Predicted

0.0

GAIAPredicted

14.7 / 100

LegalBenchPredicted

72.8 / 100

LiveBench CodingPredicted

67.0 / 100

LiveBench Data AnalysisPredicted

44.8 / 100

LiveBench LanguagePredicted

55.3 / 100

LiveBench MathPredicted

63.7 / 100

LiveBench OverallPredicted

47.7 / 100

LiveBench ReasoningPredicted

42.4 / 100

LongBench v2 EasyPredicted

39.2

LongBench v2 HardPredicted

36.4

LongBench v2Predicted

33.1 / 100

MathVistaPredicted

52.5 / 100

MedQA (USMLE)Predicted

71.0

MLE-benchPredicted

22.2 / 100

MMMUPredicted

61.5 / 100

MMTU Table UnderstandingPredicted

48.6 / 100

MT-BenchPredicted

7.8 / 10

NoLiMa (NIAH)Predicted

90.4 / 100

OCRBench v2Predicted

74.7 / 100

Open LLM AveragePredicted

35.1 / 100

Open LLM: BBHPredicted

61.4 / 100

Open LLM: GPQAPredicted

33.0 / 100

Open LLM: IFEvalPredicted

70.7 / 100

Open LLM: MATH Level 5Predicted

31.2 / 100

Open LLM: MMLU-PROPredicted

45.8 / 100

Open LLM: MUSRPredicted

42.9 / 100

RE-BenchPredicted

4.1

SimpleBenchPredicted

24.4 / 100

simpleqaPredicted

22.0

SWE-bench LitePredicted

12.4 / 100

SWE-bench VerifiedPredicted

35.7 / 100

tau-bench RetailPredicted

79.7 / 100

Terminal-Bench HardPredicted

0.0 / 100

Vectara Factual ConsistencyPredicted

92.9 / 100

Vectara Hallucination RatePredicted

7.1 / 100

WebArenaPredicted

2.4 / 100

WeirdMLPredicted

20.5 / 100

WildBenchPredicted

45.7

Epoch Capabilities Indexepoch_ai

1.152e+24