FrontierMath v2 · Tiers 1–3
Epoch AI / Reasoning · Research-level mathematical reasoning
Across e/accCounts toward ranking
Evidence budget3.6%
Upstream data time10/16 02:52
Last successful sync10/17 08:05
What it measures and how
Use only v2 Tiers 1–3. Do not mix older tasks or Tier 4 into this metric.
How this evidence is used
Counts toward ranking; benchmarks in the same task family share a fixed budget.
Benchmark results
Fixed rules select one representative configuration per public model. Anonymous test models are omitted. Original source ranks are retained, with up to 30 models per benchmark.
| Source rank | Source model name | Raw result | Representative configuration |
|---|---|---|---|
| 1 | gpt-6.1-sol_maxOpenAI | 93.7% | Max reasoning |
| 1 | gpt-6-astra_maxOpenAI | 93.7% | Max reasoning |
| 3 | claude-opus-5-5_maxAnthropic | 91.2% | Max reasoning |
| 4 | claude-fable-5-1_maxAnthropic | 90.2% | Max reasoning |
| 5 | gpt-6-sol_maxOpenAI | 89.8% | Max reasoning |
| 6 | gpt-5.6-sol_maxOpenAI | 89.1% | Max reasoning |
| 7 | claude-sonnet-5-5_maxAnthropic | 88.8% | Max reasoning |
| 8 | gpt-5.5-pro_xhighOpenAI | 87.7% | xHigh reasoning |
| 9 | claude-fable-5_maxAnthropic | 87.0% | Max reasoning |
| 10 | gpt-5.6-terra_maxOpenAI | 86.0% | Max reasoning |
| 10 | gemini-4-argon_highGoogle | 86.0% | High reasoning |
| 12 | claude-opus-5_maxAnthropic | 85.6% | Max reasoning |
| 13 | gpt-5.5_xhighOpenAI | 85.3% | xHigh reasoning |
| 14 | gpt-5.4-pro-2026-03-05_xhighOpenAI | 82.5% | xHigh reasoning |
| 15 | gpt-5.6-luna_maxOpenAI | 82.1% | Max reasoning |
| 15 | muse-wonder-1.0_maxMeta AI | 82.1% | Max reasoning |
| 17 | claude-opus-4-8_maxAnthropic | 80.0% | Max reasoning |
| 18 | gpt-6-luna_maxOpenAI | 78.9% | Max reasoning |
| 19 | gpt-5.4-2026-03-05_xhighOpenAI | 78.6% | xHigh reasoning |
| 20 | claude-haiku-5-5_maxAnthropic | 75.1% | Max reasoning |
| 21 | qwen3.8-max_xhighAlibaba | 74.7% | xHigh reasoning |
| 23 | muse-spark-1.3_maxMeta | 74.0% | Max reasoning |
| 24 | gpt-5.2-pro-2025-12-11_xhighOpenAI | 74.0% | xHigh reasoning |
| 25 | kimi-k3_maxMoonshot AI | 72.2% | Max reasoning |
| 26 | gemini-3.7-flash_highGoogle | 71.6% | High reasoning |
| 27 | claude-opus-4-7_maxAnthropic | 70.2% | Max reasoning |
| 28 | Muse-Spark-Open-816B-A42B_xhighMeta AI | 69.8% | xHigh reasoning |
| 29 | glm-5.3_maxZ.ai | 68.8% | Max reasoning |
| 30 | gemini-3.8-flash_highGoogle | 68.4% | High reasoning |
| 31 | gpt-5.2-2025-12-11_xhighOpenAI | 67.4% | xHigh reasoning |
Limitations and attribution
Counts toward ranking; benchmarks in the same task family share a fixed budget.
Data license: CC BY 4.0 · Data evaluated by Epoch AI itself
Results are published by Epoch AI. Raw scores and the e/acc consensus index use different scales and cannot be added directly.