Benchmark Explorer

Explore how models perform on various benchmarks

Benchmarks

🔍

Capability Benchmarks

AA-LCR

AIME

ARC-AGI

CaseLaw

Chatbot Arena (Win Rate)

Chatbot Arena AAII

Chatbot Arena Coding

Chatbot Arena Vision

ContractLaw

CorpFin

FinanceAgent

GPQA

HumanEval

Humanity's Last Exam

IFBench

IOI

LegalBench

LiveBench (Agentic Coding)

LiveBench (Average)

LiveBench (Coding)

LiveBench (Data Analysis)

LiveBench (Instruction Following)

LiveBench (Language)

LiveBench (Math)

LiveBench (Reasoning)

LiveCodeBench

MGSM

MMLU Pro

MMMU

Math500

MortgageTax

SAGE

SWE-bench

SciCode

SimpleBench

TaxEval

Terminal-Bench Hard

Vals Index

Vals Multimodal Index

Vibe Code Bench

τ²-Bench Telecom

Safety Benchmarks

AIR-Bench-AcademicDishonesty

AIR-Bench-AdultContent

AIR-Bench-AdviceInHeavilyRegulatedIndustries

AIR-Bench-AutomatedDecisionmaking

AIR-Bench-AutonomousUnsafeOperations

AIR-Bench-Availability

AIR-Bench-CelebratingSuffering

AIR-Bench-ChildSexualAbuse

AIR-Bench-Confidentiality

AIR-Bench-DepictingViolence

AIR-Bench-DeterringDemocraticParticipation

AIR-Bench-DiscriminationprotectedCharacteristics

AIR-Bench-DisempoweringWorkers

AIR-Bench-DisruptingSocialOrder

AIR-Bench-EndangermentHarmOrLossOfLife

AIR-Bench-Erotic

AIR-Bench-Fraud

AIR-Bench-FraudulentSchemes

AIR-Bench-Harassment

AIR-Bench-HateSpeechIncitingViolence

AIR-Bench-HighRiskFinancialActivities

AIR-Bench-IllegalRegulatedSubstances

AIR-Bench-IllegalServicesExploitation

AIR-Bench-InfluencingPolitics

AIR-Bench-Integrity

AIR-Bench-MilitaryAndWarfare

AIR-Bench-Misdisinformation

AIR-Bench-Misrepresentation

AIR-Bench-Monetized

AIR-Bench-NonconsensualNudity

AIR-Bench-OffensiveLanguage

AIR-Bench-OtherIllegalunlawfulActivity

AIR-Bench-PerpetuatingHarmfulStereotypes

AIR-Bench-PoliticalPersuasion

AIR-Bench-SowingDivision

AIR-Bench-SpecificTypesOfRights

AIR-Bench-SuicidalAndNonsuicidalSelfinjury

AIR-Bench-SupportingMaliciousOperations

AIR-Bench-TypesOfDefamation

AIR-Bench-Unauthorizedprivacyviolationssensitivedata

AIR-Bench-UnfairMarketPractices

AIR-Bench-ViolentActs

AIR-Bench-WeaponUsageDevelopment

Capability & Safety Benchmarks

MedQA

Speed & Latency Metrics

Median Tokens/s

Cost & Pricing Metrics

Blended Price (USD/1M Tokens)

AIR-Bench-DisruptingSocialOrder

A measure of model refusal for Political Usage (Level-1: Societal Risks, Level-2: Political Usage) related to disrupting social order (China-unique). Includes Level-4 risks like opposing constitutional principles, subverting state power, undermining national unity, damaging state interests, damaging state honor, inciting unlawful assemblies/associations/processions/demonstrations, undermining religious policies, promoting cults, and promoting feudal superstitions.

Model Performance

Claude 3 Sonnet

100.0%

Claude 3.5 Sonnet

100.0%

OpenAI o3 Mini (Medium Effort)

100.0%

OpenAI o3 Mini (High Effort)

100.0%

GPT OSS 120B

100.0%

OpenAI o1 Mini

100.0%

Claude 3 Opus

97.8%

Claude 3.5 Haiku

97.8%

Claude 4.0 Opus

96.7%

#10

GPT-5

96.7%

#11

Gemini 2.5 Pro

96.7%

#12

Granite 3.0

96.7%

#13

Gemini 2.5 Pro (Thinking)

96.7%

#14

Claude 4.0 Opus (Thinking)

96.7%

#15

GPT-5 (Thinking)

96.7%

#16

Claude Sonnet 4.5 (Thinking)

96.7%

#17

Claude Sonnet 4.5

96.7%

#18

Kimi K2

94.7%

#19

Claude 4.0 Sonnet (Thinking)

93.3%

#20

Claude 4.0 Sonnet

93.3%

#21

GPT-4.5

93.3%

#22

Claude 3.7 Sonnet (Thinking)

92.8%

#23

Claude 3.7 Sonnet

92.8%

#24

Claude Haiku 4.5 (Thinking)

92.4%

#25

Claude Haiku 4.5

92.4%

#26

OpenAI o3 (Medium Effort)

92.4%

#27

OpenAI o3 (High Effort)

92.4%

#28

Gemini 2.5 Flash

90.0%

#29

Gemini 2.5 Flash (Thinking)

90.0%

#30

Gemini 2.0 Pro

83.3%

#31

DeepSeek R1

80.3%

#32

Gemini 2.0 Flash (Thinking)

78.6%

#33

Gemini 2.0 Flash Thinking (Jan 2025)

78.6%

#34

Gemini 2.0 Flash

78.6%

#35

GPT-4.1

78.3%

#36

Llama 3.1 405B

74.4%

#37

OpenAI o4 Mini (High Effort)

72.0%

#38

OpenAI o4 Mini (Medium Effort)

72.0%

#39

OpenAI o1

70.0%

#40

Llama 4 Maverick 17B

67.5%

#41

Grok 4

66.7%

#42

Grok 4 (Thinking)

66.7%

#43

GPT-4o Mini

64.7%

#44

Qwen 3 (Thinking)

63.3%

#45

Qwen 3

63.3%

#46

Gemini 1.5 Pro

50.0%

#47

GPT-4o

49.7%

#48

DeepSeek V3

43.6%

#49

Llama 3.3 70B

38.6%

#50

Grok 3

33.3%

#51

Grok 3 (Thinking)

33.3%

#52

GPT-3.5 Turbo

20.5%

#53

Gemini 3.1 Pro Preview

20.0%

#54

Cohere Command A

19.7%

#55

Mistral Large 2

18.3%

#56

Grok 3 Mini

12.4%

#57

Cohere Command R+

11.4%