llmboard.aiAI model intelligence
Home

Model Rankings

OverallOpen ModelsAgentCodingReasoningMathKnowledgeInstruction FollowingTextVision
Image GenerationImage Editing
Video GenerationImage to VideoVideo Editing
Text to SpeechSpeech to Text
Embeddings

Efficiency

Chat Token PricingImage PricingVideo PricingAudio Pricing
Chat Speed & LatencyProvider Reliability

Benchmarks

GPQAMMLU-ProAIME 2025SWE-Bench VerifiedMMLUHumanity's Last ExamLiveCodeBenchMATHHumanEvalMMMU-Pro
All Benchmarks

Tools

Model DirectoryCompare Models

Scoring & Data

Scoring & Data
393 models668 benchmarks

Leaderboard Center

Overall RankingCodingCore BenchmarksPrice & ValueRuntime Performance

Modalities

All ModelsImage GenerationImage EditingVideo GenerationImage-to-VideoVideo EditingText-to-SpeechSpeech-to-TextEmbeddings

Data & Methods

Scoring MethodAll BenchmarksReasoningMath

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai

math benchmark

MATH-500 Leaderboard

MATH-500 is a subset of the MATH dataset containing 500 challenging competition mathematics problems from AMC 10, AMC 12, AIME, and other mathematics competitions. Each problem includes full step-by-step solutions and spans multiple difficulty levels across seven mathematical subjects including Prealgebra, Algebra, Number Theory, Counting and Probability, Geometry, Intermediate Algebra, and Precalculus.

Updated Aug 17, 2026

Models32
Model coverage32
MetricScore
EvidenceC

On this page

  • Ranking
  • Highlights
  • Distribution
  • Top models
  • About
  • FAQ

MATH-500 Ranking

Higher score ranks better on this benchmark.

30 of 32 rows
Columns

Show columns

Sort by
Rank
Model
Score
Percentile
Participants
Evidence
Evaluated
Rank01ModelMELongCat-Flash-ThinkingMeituanScore99.2%Percentile100.0%Participants32EvidenceCEvaluatedAug 17, 2026
Rank02ModelSASarvam-105BSarvam AIScore98.6%Percentile96.8%Participants32EvidenceCEvaluatedAug 17, 2026
Rank03ModelZAGLM-4.5Zhipu AIScore98.2%Percentile93.5%Participants32EvidenceCEvaluatedAug 17, 2026
Rank04ModelZAGLM-4.5-AirZhipu AIScore98.1%Percentile90.3%Participants32EvidenceCEvaluatedAug 17, 2026
Rank05ModelNVNemotron Nano 9B v2NVIDIAScore97.8%Percentile87.1%Participants32EvidenceCEvaluatedAug 17, 2026
Rank06ModelMAKimi K2 InstructMoonshot AIScore97.4%Percentile83.9%Participants32EvidenceCEvaluatedAug 17, 2026
Rank07ModelMAKimi K2-Instruct-0905Moonshot AIScore97.4%Percentile80.7%Participants32EvidenceCEvaluatedAug 17, 2026
Rank08ModelNVLlama 3.1 Nemotron Ultra 253B v1NVIDIAScore97.0%Percentile77.4%Participants32EvidenceCEvaluatedAug 17, 2026
Rank09ModelSASarvam-30BSarvam AIScore97.0%Percentile74.2%Participants32EvidenceCEvaluatedAug 17, 2026
Rank10ModelMELongCat-Flash-LiteMeituanScore96.8%Percentile71.0%Participants32EvidenceCEvaluatedAug 17, 2026
Rank11ModelMIMiniMax M1 80KMiniMaxScore96.8%Percentile67.7%Participants32EvidenceCEvaluatedAug 17, 2026
Rank12ModelNVLlama-3.3 Nemotron Super 49B v1NVIDIAScore96.6%Percentile64.5%Participants32EvidenceCEvaluatedAug 17, 2026
Rank13ModelMELongCat-Flash-ChatMeituanScore96.4%Percentile61.3%Participants32EvidenceCEvaluatedAug 17, 2026
Rank14ModelANClaude 3.7 SonnetAnthropicScore96.2%Percentile58.1%Participants32EvidenceCEvaluatedAug 17, 2026
Rank15ModelMAKimi-k1.5Moonshot AIScore96.2%Percentile54.8%Participants32EvidenceCEvaluatedAug 17, 2026
Rank16ModelMIMiniMax M1 40KMiniMaxScore96.0%Percentile51.6%Participants32EvidenceCEvaluatedAug 17, 2026
Rank17ModelDEDeepSeek R1 ZeroDeepSeekScore95.9%Percentile48.4%Participants32EvidenceCEvaluatedAug 17, 2026
Rank18ModelNVLlama 3.1 Nemotron Nano 8B V1NVIDIAScore95.4%Percentile45.2%Participants32EvidenceCEvaluatedAug 17, 2026
Rank19ModelMIPhi 4 Mini ReasoningMicrosoftScore94.6%Percentile41.9%Participants32EvidenceCEvaluatedAug 17, 2026
Rank20ModelDEDeepSeek R1 Distill Llama 70BDeepSeekScore94.5%Percentile38.7%Participants32EvidenceCEvaluatedAug 17, 2026
Rank21ModelDEDeepSeek R1 Distill Qwen 32BDeepSeekScore94.3%Percentile35.5%Participants32EvidenceCEvaluatedAug 17, 2026
Rank22ModelDEDeepSeek-V3 0324DeepSeekScore94.0%Percentile32.3%Participants32EvidenceCEvaluatedAug 17, 2026
Rank23ModelDEDeepSeek R1 Distill Qwen 14BDeepSeekScore93.9%Percentile29.0%Participants32EvidenceCEvaluatedAug 17, 2026
Rank24ModelDEDeepSeek R1 Distill Qwen 7BDeepSeekScore92.8%Percentile25.8%Participants32EvidenceCEvaluatedAug 17, 2026
Rank25ModelACQwQ-32BAlibaba Cloud / Qwen TeamScore90.6%Percentile22.6%Participants32EvidenceCEvaluatedAug 17, 2026
Rank26ModelACQwQ-32B-PreviewAlibaba Cloud / Qwen TeamScore90.6%Percentile19.4%Participants32EvidenceCEvaluatedAug 17, 2026
Rank27ModelDEDeepSeek-V3DeepSeekScore90.2%Percentile16.1%Participants32EvidenceCEvaluatedAug 17, 2026
Rank28ModelOPo1-miniOpenAIScore90.0%Percentile12.9%Participants32EvidenceCEvaluatedAug 17, 2026
Rank29ModelDEDeepSeek R1 Distill Llama 8BDeepSeekScore89.1%Percentile9.7%Participants32EvidenceCEvaluatedAug 17, 2026
Rank30ModelDEDeepSeek R1 Distill Qwen 1.5BDeepSeekScore83.9%Percentile6.5%Participants32EvidenceCEvaluatedAug 17, 2026
Rank31ModelIBGranite 3.3 8B BaseIBMScore69.0%Percentile3.2%Participants32EvidenceCEvaluatedAug 17, 2026
Rank32ModelIBGranite 3.3 8B InstructIBMScore69.0%Percentile0.0%Participants32EvidenceCEvaluatedAug 17, 2026

MATH-500 Highlights

The leading models and scores on this benchmark.

Rank #1LongCat-Flash-Thinking99.2%Rank #2Sarvam-105B98.6%Rank #3GLM-4.598.2%Rank #4GLM-4.5-Air98.1%

MATH-500 Score Distribution

A closer view of the leading scores on this benchmark.

MATH-500

The Top AI Models for MATH-500

The first five results on this benchmark, with official price and output speed added where the model identity can be matched.

Ranking basisThis math-500 AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.

  1. 01
    ME
    LongCat-Flash-ThinkingMeituan
    Score
    99.2%
    Speed
    Up to 100 tok/s via Meituan

    Strengths

    • Ranks #1 of 32 compared models
    • 100th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MATH-500, not total model capability
  2. 02
    SA
    Sarvam-105BSarvam AI
    Score
    98.6%

    Strengths

    • Ranks #2 of 32 compared models
    • 97th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MATH-500, not total model capability
  3. 03
    ZA
    GLM-4.5Zhipu AI
    Score
    98.2%
    Price
    $0.60 input / $2.2 output per 1M tokens

    Strengths

    • Ranks #3 of 32 compared models
    • 94th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MATH-500, not total model capability
  4. 04
    ZA
    GLM-4.5-AirZhipu AI
    Score
    98.1%
    Price
    $0.20 input / $1.1 output per 1M tokens

    Strengths

    • Ranks #4 of 32 compared models
    • 90th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MATH-500, not total model capability
  5. 05
    NV
    Nemotron Nano 9B v2NVIDIA
    Score
    97.8%

    Strengths

    • Ranks #5 of 32 compared models
    • 87th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MATH-500, not total model capability

Selection summary

Best AI Models for MATH-500

LongCat-Flash-Thinking currently leads MATH-500 with 99.2%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.

Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.

Benchmark rank #1LongCat-Flash-Thinking99.2% · Up to 100 tok/s via MeituanBenchmark rank #2Sarvam-105B98.6%Benchmark rank #3GLM-4.598.2% · $0.60 input / $2.2 output per 1M tokens

What is MATH-500?

What MATH-500 measures and how its scores work.

MATH-500 is a subset of the MATH dataset containing 500 challenging competition mathematics problems from AMC 10, AMC 12, AIME, and other mathematics competitions. Each problem includes full step-by-step solutions and spans multiple difficulty levels across seven mathematical subjects including Prealgebra, Algebra, Number Theory, Counting and Probability, Geometry, Intermediate Algebra, and Precalculus.

Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of C.

Family
MATH-500
Modality
text
Primary category
math
Score direction
higher
LLMBoard eligible
Yes
Evaluation key
math-500|llm-stats-current

Benchmark scores retain their original unit. Overall score eligibility is shown separately.

FAQ

Common questions about MATH-500.

Which model scores highest on MATH-500?

LongCat-Flash-Thinking is currently ranked first with 99.2%.

What does MATH-500 measure?

MATH-500 is a subset of the MATH dataset containing 500 challenging competition mathematics problems from AMC 10, AMC 12, AIME, and other mathematics competitions. Each problem includes full step-by-step solutions and spans multiple difficulty levels across seven mathematical subjects including Prealgebra, Algebra, Number Theory, Counting and Probability, Geometry, Intermediate Algebra, and Precalculus.

Is a higher score better?

Yes. Higher values rank better for this benchmark.

How many models are compared?

32 model results are currently shown.

Does this benchmark affect the overall score?

Yes. This benchmark can contribute to the current LLMBoard capability score.