Epoch AI launches FrontierMath Erdős benchmark; Astra solves 2 of 68 unsolved problems
littmath · x · 2026-09-04
Epoch AI and mathematician Thomas Bloom launched FrontierMath Erdős: 68 curated unsolved Erdős problems where AI must write solutions in Lean for automated verification. It responds to Terence Tao's ICM remark that AI math evidence is "a confusing mess" gathered without proper scientific conditions. No prior model solved any problem; GPT-6 Astra solved 2/68 (3%). An open-source scaffold is available.
More from Models
- Early Access Tester: GPT-6 Astra Built a Blender Werewolf in 8 Minutes and a World Simulator in 17 — TheMoonMidas · 2026-09-04
- swyx Burned 20B Tokens Stress-Testing Astra on Real AI Engineering Tasks — All for Under $6/Hour — charliermarsh · 2026-09-04
- OpenAI Launches GPT-6 Astra: 88% on Reverse-Engineering Benchmark, 1M Context Accuracy Jumps to 96.3% — rickasaurus · 2026-09-04
- GPT-6 Astra launch called a nothingburger: limited preview gates most users for weeks — ns123abc · 2026-09-04
- GPT-6 Astra priced at $10/$50, 2.5x GPT-5.6 Sol, matches Fable 5 coding agents at half cost — cedric_chee · 2026-09-04
- IFM's 36B MoE model K2-Horizon-MoVA-36B-A4B trends on Hugging Face — IFM · 2026-09-04