Scoreboard of LLM open-problem solves in math
ctjlewis · x · 2026-07-21
A scoreboard claims that, as of July 2026, more than 20 named open math problems have been credited to U.S. labs or U.S.-based prompters via LLMs, while China has 0 credited solves and 1 partial contribution.
The chart below breaks the credited results down by lab: DeepMind leads with 9, followed by OpenAI with 6, Axiom with 4, Harmonic with 3, Anthropic with 1, ByteDance with about 1, and DeepSeek/Alibaba/Moonshot/Zhipu at 0. The note says the count includes named open problems where the model is credited as solver or co-solver, and excludes the 100 Erdős problems moved to “solved” since Oct. 2025 because most were literature retrieval rather than novel mathematics.
Related event: LLM Math Benchmark Sparks Debate(2 posts)→
More from Research
- New paper defines self-state attacks, showing OS defenses leave four agent-memory cases indistinguishable — Justgototheeffinmoon · 2026-07-22
- Animation shows how an MLP’s first-layer weights change while learning MNIST — CatAstro_Piyush · 2026-07-22
- Project APE finds verifier reliability drops when papers contain multiple errors — soumitrashukla9 · 2026-07-22
- Project APE says verifier costs fell about 90x in a year as Chinese open models lead — soumitrashukla9 · 2026-07-22
- OpenAI-linked paper says capability RL can make models more reward-seeking — MariusHobbhahn · 2026-07-22
- Project APE builds its verifier benchmark from 100 AI-written papers with injected errors — soumitrashukla9 · 2026-07-22