Benchmark
AIME 2025
Math benchmarks measure how well models solve maths problems. Scores are percentages. Higher is better.
- Models measured
- 104
- Scores recorded
- 147
- Sources
- 2
- IQ weight
- 1.67
Score distribution
- 81.5%
- Mean
- 88%
- Median
- 19.2%
- Std dev
- 51.9%
- 10th percentile
- 99.6%
- 90th percentile
- 6.7%
- Min
This benchmark has a weight of 1.67 for chat models, measured over 104 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 51 | Claude Opus 4.5no tools · 64K thinking · 200K context · avg@5 · contamination caveat· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 92.8% |
| 52 | o4 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 92.7% |
| 53 | GPT OSS 120Bhigh· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 92.5% |
| 54 | GPT OSS 120Bofficial model card Table 3 · reasoning high · no tools · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 92.5% |
| 55 | Qwen 3 235B A22B Thinking 2507· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 92.3% |
| 56 | Grok 4 Fast· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 92% |
| 57 | GPT OSS 20bofficial model card Table 3 · reasoning high · no tools · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91.7% |
| 58 | Grok 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91.7% |
| 59 | GLM 4.7 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91.6% |
| 60 | GPT OSS 120Bofficial model card Table 3 · reasoning medium · with tools; Table 3 does not name the exact tool · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91.6% |
| 61 | GPT-5 mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91.1% |
| 62 | Mercury 2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91.1% |
| 63 | Nova 2 Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91% |
| 64 | Nova 2 LiteAmazon Nova 2 Lite official launch chart / technical report · accuracy, v1· Reported by Amazon Web Services, Amazon AGI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91% |
| 65 | Grok 3 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 90.8% |
| 66 | GPT OSS 20bofficial model card Table 3 · reasoning medium · with tools; Table 3 does not name the exact tool · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 90.4% |
| 67 | Nemotron 3 Super 120B A12B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 90.2% |
| 68 | Nemotron 3 Super 120B A12BNeMo Evaluator / mostly NeMo Skills · no tools· Reported by NVIDIA· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 90.2% |
| 69 | Claude Opus 4extended thinking · parallel test-time compute · top_p 0.95· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 90% |
| 70 | Command A Plus 05 2026· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 90% |
| 71 | Qwen3 VL 235Bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 89.7% |
| 72 | DeepSeek V3.2 Exp· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 89.3% |
| 73 | Granite 4.2 30B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 89.2% |
| 74 | Gemini 2.5 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 88% |
| 75 | Qwen3 Next 80B A3Bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 87.8% |
| 76 | DeepSeek R1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 87.5% |
| 77 | Claude Sonnet 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 87% |
| 78 | Claude Sonnet 4.5no tools· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 87% |
| 79 | ERNIE 5.0· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 87% |
| 80 | Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 87% |
| 81 | Granite 4.2 8B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 86.7% |
| 82 | OpenAI o3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 86.4% |
| 83 | MiniMax M2.5official appendix· Reported by MiniMax· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 86.3% |
| 84 | Mistral Medium 3.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 86.3% |
| 85 | GPT-5 nano· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 85.2% |
| 86 | Claude Sonnet 4extended thinking · parallel test-time compute · top_p 0.95· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 85% |
| 87 | Ministral 14Blatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 85% |
| 88 | Mistral Smalllatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 83.8% |
| 89 | Qwen3 VL 32B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 83.7% |
| 90 | qwen3-vl-30bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 83.1% |
| 91 | Qwen3 Max· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.6% |
| 92 | Qwen 3 235B A22B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.5% |
| 93 | GPT-5.5 Instant· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81.2% |
| 94 | MiniMax M2.1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 81% |
| 95 | Claude Haiku 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 80.7% |
| 96 | Claude Haiku 4.5no tools· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.7% |
| 97 | Qwen: Qwen3 VL 8B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. | 80.3% |
| 98 | GPT OSS 120Bofficial model card Table 3 · reasoning medium · no tools · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80% |
| 99 | Ministral 8Blatest· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 78.7% |
| 100 | Granite 4.2 3B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 78.3% |
Showing 51–100 of 147 results.
Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats