Benchmarks / Math
FrontierMath Erdős
Epoch AI's 68 open Erdős problems, formulated in Lean. Models must supply complete formal proofs or disproofs. Published solved fraction; distinct from FrontierMath difficulty tiers. Data: Epoch AI, CC BY.
FrontierMath Erdős
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | GPT-6 Astra | max | 2.9% |
| 2 | Opus 5.5 | max | 2.9% |
| 3 | Sonnet 5.5 | max | 2.9% |
| 4 | Fable 5.1 | max | 0.0% |
| 5 | GPT-5.5 | xhigh | 0.0% |
| 6 | Fable 5 | max | 0.0% |
| 7 | GPT-5.6 Sol | max | 0.0% |