Skip to content
Trending storyMonitoring

Tsinghua and Moonshot AI release TCSExam, a 100-task theoretical computer science benchmark

1 article1 sourcesince Nov 8Last article

Overview

AISummary of 1 article

Tsinghua University and Moonshot AI release TCSExam, a benchmark of 100 expert-reviewed tasks across four tracks: Lean theorem proving, solver engineering, record-setting constructions, and 30 open conjectures.

Without supplied lemmas, GPT-6 Astra fully proved theorems in 70.0% of attempts and Kimi K3.1 Pro in 38.3%, according to the report.

On the 30 open conjectures, the same models resolved none. The few accepted conjecture submissions adapted counterexamples and proofs that OpenAI had already published.

Written by AI from the articles below · updated Nov 9, 9:25 PM ET

Check the sources:

Latest developmentNov 8, 1:45 PM ET
Tsinghua and Moonshot AI release TCSExam, a theoretical computer science research benchmark

Article timeline

The articles in this story. Times are ET.

Nov 8
  1. Rohan PaulX
    Tsinghua and Moonshot AI release TCSExam, a theoretical computer science research benchmark

    AITsinghua University and Moonshot AI introduce TCSExam, a benchmark of 100 expert-reviewed tasks across Lean theorem proving, solver engineering, record-setting constructions, and 30 open conjectures. Without supplied lemmas, GPT-6 Astra fully proved theorems in 70.0% of attempts, and Kimi K3.1 Pro in 38.3%. The same model resolved 0 of 30 open conjectures, and the few accepted conjecture submissions adapted counterexamples and proofs OpenAI had already published.

    Image from @rohanpaul_ai's post

Heat Trend

Not enough continuous observations to show a trend yet.