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
1367 models808 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

AA AIME 2025 Leaderboard

AIME I + II; 30 questions; 10 runs per question; pass@1; historical results

Updated Sep 22, 2026

Models96
Model coverage96
Metricaime25
EvidenceB

On this page

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

AA AIME 2025 Ranking

Higher aime25 ranks better on this benchmark.

30 of 96 rows
Columns

Show columns

Sort by
Rank
Model
aime25
Percentile
Participants
Evidence
Evaluated
Rank01ModelOPGPT-5 CodexOpenAIaime2598.67%Percentile100.00%Participants96EvidenceBEvaluatedN/A
Rank02ModelGOGemini 3 FlashGoogleaime2597.00%Percentile98.95%Participants96EvidenceBEvaluatedN/A
Rank03ModelDEDeepSeek-V3.2-SpecialeDeepSeekaime2596.67%Percentile97.89%Participants96EvidenceBEvaluatedN/A
Rank04ModelXIMiMo-V2-FlashXiaomiaime2596.33%Percentile96.84%Participants96EvidenceBEvaluatedN/A
Rank05ModelOPGPT-5.1 CodexOpenAIaime2595.67%Percentile95.79%Participants96EvidenceBEvaluatedN/A
Rank06ModelGOGemini 3 ProGoogleaime2595.67%Percentile94.74%Participants96EvidenceBEvaluatedN/A
Rank07ModelZAGLM-4.7Zhipu AIaime2595.00%Percentile93.68%Participants96EvidenceBEvaluatedN/A
Rank08ModelMAKimi K2-Thinking-0905Moonshot AIaime2594.67%Percentile92.63%Participants96EvidenceBEvaluatedN/A
Rank09ModelOPGPT-5OpenAIaime2594.33%Percentile91.58%Participants96EvidenceBEvaluatedN/A
Rank10ModelOPGPT-5.1OpenAIaime2594.00%Percentile90.53%Participants96EvidenceBEvaluatedN/A
Rank11ModelOPGPT OSS 120BOpenAIaime2593.44%Percentile89.47%Participants96EvidenceBEvaluatedN/A
Rank12ModelXAGrok-4xAIaime2592.67%Percentile88.42%Participants96EvidenceBEvaluatedN/A
Rank13ModelDEDeepSeek-V3.2DeepSeekaime2592.00%Percentile87.37%Participants96EvidenceBEvaluatedN/A
Rank14ModelOPGPT-5 MediumOpenAIaime2591.67%Percentile86.32%Participants96EvidenceBEvaluatedN/A
Rank15ModelOPGPT-5.1 Codex MiniOpenAIaime2591.67%Percentile85.26%Participants96EvidenceBEvaluatedN/A
Rank16ModelANClaude Opus 4.5Anthropicaime2591.33%Percentile84.21%Participants96EvidenceBEvaluatedN/A
Rank17ModelNVNemotron 3 Nano (30B A3B)NVIDIAaime2591.00%Percentile83.16%Participants96EvidenceBEvaluatedN/A
Rank18ModelACQwen3-235B-A22B-Instruct-2507Alibaba Cloud / Qwen Teamaime2591.00%Percentile82.11%Participants96EvidenceBEvaluatedN/A
Rank19ModelOPGPT-5 miniOpenAIaime2590.67%Percentile81.05%Participants96EvidenceBEvaluatedN/A
Rank20ModelOPo4-miniOpenAIaime2590.67%Percentile80.00%Participants96EvidenceBEvaluatedN/A
Rank21ModelDEDeepSeek-V3.1DeepSeekaime2589.67%Percentile78.95%Participants96EvidenceBEvaluatedN/A
Rank22ModelXAGrok-4 Fast ReasoningxAIaime2589.67%Percentile77.89%Participants96EvidenceBEvaluatedN/A
Rank23ModelOPGPT OSS 20BOpenAIaime2589.33%Percentile76.84%Participants96EvidenceBEvaluatedN/A
Rank24ModelXAGrok-4.1 Fast ReasoningxAIaime2589.33%Percentile75.79%Participants96EvidenceBEvaluatedN/A
Rank25ModelOPo3OpenAIaime2588.33%Percentile74.74%Participants96EvidenceBEvaluatedN/A
Rank26ModelANClaude Sonnet 4.5Anthropicaime2588.00%Percentile73.68%Participants96EvidenceBEvaluatedN/A
Rank27ModelGOGemini 2.5 ProGoogleaime2587.67%Percentile72.63%Participants96EvidenceBEvaluatedN/A
Rank28ModelZAGLM-4.6Zhipu AIaime2586.00%Percentile71.58%Participants96EvidenceBEvaluatedN/A
Rank29ModelBAERNIE 5.0Baiduaime2585.00%Percentile70.53%Participants96EvidenceBEvaluatedN/A
Rank30ModelOPGPT-5 nanoOpenAIaime2583.67%Percentile69.47%Participants96EvidenceBEvaluatedN/A

AA AIME 2025 Highlights

The leading models and scores on this benchmark.

AA AIME 2025 Score Distribution

A closer view of the leading scores on this benchmark.

AA AIME 2025

The Top AI Models for AA AIME 2025

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

What is AA AIME 2025?

What AA AIME 2025 measures and how its scores work.

AIME I + II; 30 questions; 10 runs per question; pass@1; historical results

Scores are shown in ratio. This benchmark is verified and has an evidence level of B.

Family
aime
Modality
text
Primary category
math
Score direction
higher
LLMBoard eligible
Yes
Evaluation key
overall

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

FAQ

Common questions about AA AIME 2025.

Which model scores highest on AA AIME 2025?

GPT-5 Codex is currently ranked first with 98.67%.

What are the top three models on AA AIME 2025?

The current leaders are GPT-5 Codex (98.67%), Gemini 3 Flash (97.00%), and DeepSeek-V3.2-Speciale (96.67%).

Which AA AIME 2025 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 AA AIME 2025 results?

The fastest matched records are GPT OSS 20B (1,000.00 tok/s via Groq), Llama 4 Scout (776.10 tok/s via Groq), and Llama 4 Maverick (638.70 tok/s via Sambanova).

Does the highest aime25 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 AA AIME 2025 measure?

AIME I + II; 30 questions; 10 runs per question; pass@1; historical results

Is a higher aime25 better?

Yes. Higher values rank better for this benchmark.

How many models are compared?

96 model results are currently shown.

Does this benchmark affect the overall score?

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

Rank #1GPT-5 Codex98.67%
Rank #2Gemini 3 Flash97.00%
Rank #3DeepSeek-V3.2-Speciale96.67%
Rank #4MiMo-V2-Flash96.33%

Ranking basisThis aa aime 2025 AI model leaderboard uses descending aime25 in the benchmark original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.

  1. 01
    OP
    GPT-5 CodexOpenAI
    aime25
    98.67%

    Strengths

    • Ranks #1 of 96 compared models
    • 100th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures AA AIME 2025, not total model capability
  2. 02
    GO
    Gemini 3 FlashGoogle
    aime25
    97.00%
    Price
    $0.50 input / $3.0 output per 1M tokens

    Strengths

    • Ranks #2 of 96 compared models
    • 99th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures AA AIME 2025, not total model capability
  3. 03
    DE
    DeepSeek
    aime25
    96.67%
    Speed
    Up to 30.00 tok/s via DeepSeek

    Strengths

    • Ranks #3 of 96 compared models
    • 98th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures AA AIME 2025, not total model capability
  4. 04
    XI
    Xiaomi
    aime25
    96.33%
    Price
    $0.14 input / $0.28 output per 1M tokens

    Strengths

    • Ranks #4 of 96 compared models
    • 97th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures AA AIME 2025, not total model capability
  5. 05
    OP
    OpenAI
    aime25
    95.67%
    Speed
    Up to 50.00 tok/s via OpenAI

    Strengths

    • Ranks #5 of 96 compared models
    • 96th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures AA AIME 2025, not total model capability

Selection summary

Best AI Models for AA AIME 2025

GPT-5 Codex currently leads AA AIME 2025 with 98.67%. 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.

DeepSeek-V3.2-Speciale
MiMo-V2-Flash
GPT-5.1 Codex
Benchmark rank #1GPT-5 Codex98.67%
Benchmark rank #2Gemini 3 Flash97.00% · $0.50 input / $3.0 output per 1M tokens
Benchmark rank #3DeepSeek-V3.2-Speciale96.67% · Up to 30.00 tok/s via DeepSeek