Benchmark
ArxivMath
Math benchmarks measure how well models solve maths problems. Scores are percentages. Higher is better.
- Models measured
- 10
- Scores recorded
- 15
- Sources
- 2
- IQ weight
- 1.57
Score distribution
- 72.4%
- Mean
- 71.8%
- Median
- 11.7%
- Std dev
- 52.2%
- 10th percentile
- 91.2%
- 90th percentile
- 52.2%
- Min
This benchmark has a weight of 1.57 for chat models, measured over 10 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 |
|---|---|---|
| 1 | Claude Opus 5.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 91.2% |
| 2 | Claude Opus 5.5Opus 5.5 System Card capability section · August 2026 · no tools · max · 4 attempts/problem· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 91.2% |
| 3 | Claude Sonnet 5.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 86.8% |
| 4 | Claude Fable 5.1Opus 5.5 System Card capability section · August 2026 · no tools· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 82.9% |
| 5 | Claude Opus 5Opus 5.5 System Card capability section · August 2026 · no tools· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 78.1% |
| 6 | Claude Sonnet 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 72.2% |
| 7 | Claude Sonnet 5April + May 2026 · with tools · extended thinking · 4 attempts/problem· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 72.2% |
| 8 | Claude Opus 4.8Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 71.8% |
| 9 | GPT 5.5Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 71.5% |
| 10 | Hy4 preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 66.6% |
| 11 | Hy4 previewofficial model-card appendix· Reported by Tencent Hy· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 66.6% |
| 12 | Claude Sonnet 5April + May 2026 · no tools · extended thinking · 4 attempts/problem· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 65.7% |
| 13 | Gemini 3.1 Pro PreviewFable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 64.8% |
| 14 | Hy3 (Deep Infra)· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 52.2% |
| 15 | Hy3 (Deep Infra)Hy3 official Benchmark Appendix · highest reasoning tier unless noted· Reported by Tencent Hunyuan· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 52.2% |
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