Rumored Math Breakthrough by GPT-5.6
新智元 · wechat · 2026-07-15
OpenAI officially claims that GPT-5.6 Sol Ultra proved the "Circuit Double Cover Conjecture," which has puzzled the math community for 50 years, in under an hour. The task wasn't completed by a single model alone; instead, a system of 64 concurrent agents collaborated to produce the proof, ultimately formatted into a paper with the help of Codex.
The article details this system's organization: the first round required exploring multiple completely different proof paths; a "picket line" was set up to rigorously refute each candidate proof; and it mandated specific lemmas, constructions, and equations to avoid empty talk. The author attributes this breakthrough to parallel test-time compute (TTC): breaking down long-duration reasoning into a massive concurrent search, thereby compressing an exploration that might otherwise take much longer into under an hour.
Related event: GPT-5.6 reportedly cracks decades-old Erdős problems amid hype debate(10 posts)→
More from Models
- Unreleased 'GPT 6 Sol' model spotted in OpenAI's API — SteveEricJordan · 2026-09-11
- DeepSeek update keeps cache hits mid-conversation, cuts costs 36.6% — teortaxesTex · 2026-09-11
- AI Sextet offers 6 models free and unlimited for 14 days, including DeepSeek and Qwen — airesearch12 · 2026-09-11
- BullshitBench update: GPT-6-Astra beats all prior OpenAI models but still trails Anthropic — scaling01 · 2026-09-11
- Astra Scores 83% on GauntletBench, First Computer-Use Agent to Beat Human Baseline — ducha_aiki · 2026-09-11
- Kimi K2.8 Preview rolls out: near-K3 coding performance, 1M context for all tiers — teortaxesTex · 2026-09-11