清華大學與 Moonshot AI 的論文提出 TCSExam,測試 GPT-6 Astra 等模型在理論電腦科學任務中的表現;GPT-6 Astra 在 Lean 定理證明的成功率為 70.0%,但在 30 個開放猜想中一題也未解出。
AI 伴隨問答
針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:
TCSExam 同時測試 Lean 證明、求解器工程與未解猜想,結果呈現 GPT-6 Astra 在已知定理與開放問題上的表現差距。
New Tsinghua and Moonshot AI paper finds that GPT-6 Astra proves 70.0% of research-level computer science theorems in Lean, yet the same model resolves 0 of 30 open conjectures in theoretical computer science (TCS).
Theoretical computer science is a natural test for AI research, because proofs can be machine-checked and records can be beaten. Existing benchmarks cover only slices, like proofs or code.
TCSExam has 100 expert-reviewed tasks in 4 tracks: Lean theorem proving, solver engineering, record-setting constructions, and 30 open conjectures. The Lean kernel or a task checker verifies every submission.
Without supplied lemmas, GPT-6 Astra fully proved theorems in 70.0% of attempts, Kimi K3.1 Pro in 38.3%, and the other 6 agents in at most 11.7%. The few accepted conjecture submissions adapted counterexamples and proofs OpenAI had already published.