GPT-5.6 Breaks Record on Math Problem
jasondeanlee · x · 2026-07-13
Sebastien Bubeck shared a math problem he uses to track reasoning progress in models: how long a gradient flow path can be under the unit Euclidean ball constraint.
The shared post notes that the previous human record was 2.29, but @jasondeanlee pushed it to 2.28 using a "double-loop cover" prompt. Afterward, GPT-5.6 pondered for 8 hours, reducing the result to 2.26 and reportedly even writing a paper on it. Ultimately, the post showcases how model reasoning capabilities continue to advance on a single, fixed problem.
Related event: GPT-5.6 Rumored to Achieve New Breakthroughs in Math Reasoning(4 posts)→
More from Models
- Kimi K2.8 Preview rolls out: near-K3 coding performance, 1M context for all tiers — teortaxesTex · 2026-09-11
- Looking for a classifier of software engineering task shapes to pick models per task — StewartalsopIII · 2026-09-11
- DeepSeek V4 Pro API to continue after Sept 2026, billing unchanged — teortaxesTex · 2026-09-11
- DeepSeek V4.1 Flash Hits 98% of GPT-6 Astra's Score at 1.4% of the Cost in Third-Party Benchmark — ayushtweetshere · 2026-09-11
- TheZvi Polls: Has Your Coding Model Choice Changed Since Fable 5.1 and Astra? — TheZvi · 2026-09-11
- antirez Weighs In on Anthropic Banning Minors From Using Claude — antirez · 2026-09-11