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 Directory

Scoring & Data

Scoring & Data
1224 models729 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 Benchmarks
llmboard.aiCopyright 2026 llmboard.ai

math benchmark

GSM8k Leaderboard

GSM8k is a dataset of 8.5K high-quality, linguistically diverse grade school math word problems requiring multi-step reasoning and elementary arithmetic operations.

Updated Sep 7, 2026

Models48
Model coverage48
MetricScore
EvidenceB

On this page

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

GSM8k Ranking

Higher score ranks better on this benchmark.

30 of 48 rows
Columns

Show columns

Sort by
Rank
Model
Score
Percentile
Participants
Evidence
Evaluated
Rank01ModelXIMiMo-V2.5-ProXiaomiScore99.60%Percentile100.00%Participants48EvidenceCEvaluatedSep 8, 2026
Rank02ModelMAKimi K2 InstructMoonshot AIScore97.30%Percentile97.87%Participants48EvidenceCEvaluatedSep 8, 2026
Rank03ModelOPo1OpenAIScore97.10%Percentile95.74%Participants48EvidenceCEvaluatedSep 8, 2026
Rank04ModelOPGPT-4.5OpenAIScore97.00%Percentile93.62%Participants48EvidenceCEvaluatedSep 8, 2026
Rank05ModelMELlama 3.1 405B InstructMetaScore96.80%Percentile91.49%Participants48EvidenceCEvaluatedSep 8, 2026
Rank06ModelANClaude 3.5 SonnetAnthropicScore96.40%Percentile89.36%Participants48EvidenceCEvaluatedSep 8, 2026
Rank07ModelANClaude 3.5 SonnetAnthropicScore96.40%Percentile87.23%Participants48EvidenceCEvaluatedSep 8, 2026
Rank08ModelGOGemma 3 27BGoogleScore95.90%Percentile85.11%Participants48EvidenceCEvaluatedSep 8, 2026
Rank09ModelACQwen2.5 32B InstructAlibaba Cloud / Qwen TeamScore95.90%Percentile82.98%Participants48EvidenceCEvaluatedSep 8, 2026
Rank10ModelACQwen2.5 72B InstructAlibaba Cloud / Qwen TeamScore95.80%Percentile80.85%Participants48EvidenceCEvaluatedSep 8, 2026
Rank11ModelDEDeepSeek-V2.5DeepSeekScore95.10%Percentile78.72%Participants48EvidenceCEvaluatedSep 8, 2026
Rank12ModelANClaude 3 OpusAnthropicScore95.00%Percentile76.60%Participants48EvidenceCEvaluatedSep 8, 2026
Rank13ModelAMNova ProAmazonScore94.80%Percentile74.47%Participants48EvidenceCEvaluatedSep 8, 2026
Rank14ModelACQwen2.5 14B InstructAlibaba Cloud / Qwen TeamScore94.80%Percentile72.34%Participants48EvidenceCEvaluatedSep 8, 2026
Rank15ModelAMNova LiteAmazonScore94.50%Percentile70.21%Participants48EvidenceCEvaluatedSep 8, 2026
Rank16ModelGOGemma 3 12BGoogleScore94.40%Percentile68.09%Participants48EvidenceCEvaluatedSep 8, 2026
Rank17ModelACQwen3 235B A22BAlibaba Cloud / Qwen TeamScore94.39%Percentile65.96%Participants48EvidenceCEvaluatedSep 8, 2026
Rank18ModelMAMistral Large 2Mistral AIScore93.00%Percentile63.83%Participants48EvidenceCEvaluatedSep 8, 2026
Rank19ModelANClaude 3 SonnetAnthropicScore92.30%Percentile61.70%Participants48EvidenceCEvaluatedSep 8, 2026
Rank20ModelAMNova MicroAmazonScore92.30%Percentile59.57%Participants48EvidenceCEvaluatedSep 8, 2026
Rank21ModelMAKimi K2 BaseMoonshot AIScore92.10%Percentile57.45%Participants48EvidenceCEvaluatedSep 8, 2026
Rank22ModelACQwen2.5 7B InstructAlibaba Cloud / Qwen TeamScore91.60%Percentile55.32%Participants48EvidenceCEvaluatedSep 8, 2026
Rank23ModelNVLlama 3.1 Nemotron 70B InstructNVIDIAScore91.43%Percentile53.19%Participants48EvidenceCEvaluatedSep 8, 2026
Rank24ModelACQwen2.5-Coder 32B InstructAlibaba Cloud / Qwen TeamScore91.10%Percentile51.06%Participants48EvidenceCEvaluatedSep 8, 2026
Rank25ModelACQwen2 72B InstructAlibaba Cloud / Qwen TeamScore91.10%Percentile48.94%Participants48EvidenceCEvaluatedSep 8, 2026
Rank26ModelGOGemini 1.5 ProGoogleScore90.80%Percentile46.81%Participants48EvidenceCEvaluatedSep 8, 2026
Rank27ModelXAGrok-1.5xAIScore90.00%Percentile44.68%Participants48EvidenceCEvaluatedSep 8, 2026
Rank28ModelGOGemma 3 4BGoogleScore89.20%Percentile42.55%Participants48EvidenceCEvaluatedSep 8, 2026
Rank29ModelANClaude 3 HaikuAnthropicScore88.90%Percentile40.43%Participants48EvidenceCEvaluatedSep 8, 2026
Rank30ModelMIPhi-3.5-MoE-instructMicrosoftScore88.70%Percentile38.30%Participants48EvidenceCEvaluatedSep 8, 2026

GSM8k Highlights

The leading models and scores on this benchmark.

GSM8k Score Distribution

A closer view of the leading scores on this benchmark.

GSM8k

The Top AI Models for GSM8k

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

What is GSM8k?

What GSM8k measures and how its scores work.

GSM8k is a grade school math dataset containing 8.5K word problems.

It measures Score, reported as a ratio, on problems requiring multi-step reasoning and elementary arithmetic operations.

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

Family
GSM8k
Modality
text
Primary category
math
Score direction
higher
LLMBoard eligible
No
Evaluation key
overall

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

FAQ

Common questions about GSM8k.

Which model scores highest on GSM8k?

MiMo-V2.5-Pro is currently ranked first with 99.60%.

What are the top three models on GSM8k?

The current leaders are MiMo-V2.5-Pro (99.60%), Kimi K2 Instruct (97.30%), and o1 (97.10%).

Which GSM8k model has the lowest official input price?

Nova Micro has the lowest matched official input price at $0.04 input / $0.14 output per 1M tokens.

Which models are fastest among GSM8k results?

The fastest matched records are Llama 3.2 3B Instruct (171.50 tok/s via DeepInfra), Gemini 1.5 Flash (150.00 tok/s via Google), and Claude 3.5 Sonnet (100.00 tok/s via Anthropic).

Does the highest score result prove overall model quality?

No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.

What does GSM8k measure?

It measures Score, reported as a ratio, on problems requiring multi-step reasoning and elementary arithmetic operations.

Is a higher score better?

Yes. Higher values rank better for this benchmark.

How many models are compared?

48 model results are currently shown.

Does this benchmark affect the overall score?

No. This benchmark is shown for reference but does not contribute to the overall score.

Rank #1MiMo-V2.5-Pro99.60%
Rank #2Kimi K2 Instruct97.30%
Rank #3o197.10%
Rank #4GPT-4.597.00%

Ranking basisThis gsm8k AI model leaderboard uses descending score in the benchmark original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.

  1. 01
    XI
    MiMo-V2.5-ProXiaomi
    Score
    99.60%
    Price
    $0.44 input / $0.87 output per 1M tokens
    Speed
    Up to 5.27 tok/s via DeepInfra

    Strengths

    • Ranks #1 of 48 compared models
    • 100th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures GSM8k, not total model capability
  2. 02
    MA
    Kimi K2 InstructMoonshot AI
    Score
    97.30%

    Strengths

    • Ranks #2 of 48 compared models
    • 98th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures GSM8k, not total model capability
  3. 03
    OP
    OpenAI
    Score
    97.10%
    Price
    $15 input / $60 output per 1M tokens
    Speed
    Up to 66.00 tok/s via OpenAI

    Strengths

    • Ranks #3 of 48 compared models
    • 96th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures GSM8k, not total model capability
  4. 04
    OP
    OpenAI
    Score
    97.00%
    Speed
    Up to 50.00 tok/s via OpenAI

    Strengths

    • Ranks #4 of 48 compared models
    • 94th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures GSM8k, not total model capability
  5. 05
    ME
    Meta
    Score
    96.80%
    Speed
    Up to 42.00 tok/s via Google

    Strengths

    • Ranks #5 of 48 compared models
    • 91th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures GSM8k, not total model capability

Selection summary

Best AI Models for GSM8k

MiMo-V2.5-Pro currently leads GSM8k with 99.60%. 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.

o1
GPT-4.5
Llama 3.1 405B Instruct
Benchmark rank #1MiMo-V2.5-Pro99.60% · $0.44 input / $0.87 output per 1M tokens
Benchmark rank #2Kimi K2 Instruct97.30%
Benchmark rank #3o197.10% · $15 input / $60 output per 1M tokens