GPT-5.6 Sol Rumored to Solve Formal Proofs
jasondeanlee · x · 2026-07-16
The post cites a capability showcase regarding GPT-5.6 Sol: someone claimed that by using a CDC prompt and multiple sub-agents, it solved Erdős Problem #796 in a few hours and fully formalized the proof in Lean.
However, a commenter threw cold water on this, noting that trying the exact same method on a non-Erdős problem would "bankrupt you," implying this approach might not be reliable for other problems. Overall, the discussion centers on the boundaries of models performing mathematical proofs and formal verification.
Related event: GPT-5.6 Solves Long-Standing Math Problems(4 posts)→
More from Research
- Reasoning models can barely control their chain of thought, paper finds — tomekkorbak · 2026-10-27
- Eric Topol challenges frontier-model study: where's proof of sustained upskilling? — EricTopol · 2026-10-27
- New paper: LLM data schedules and learning rate schedules should be co-designed — jasondeanlee · 2026-10-27
- What makes research get adopted: important problems, proper evals, and usability — srchvrs · 2026-10-27
- Bitter lesson part 2: few academic inventions matter, and conference publications aren't the key — srchvrs · 2026-10-27
- First-week PhD student lands ARR Area Chair role, exposing peer-review quirks — anmarasovic · 2026-10-27