Today for AI
HOT RADAR
llmHEAT 9.7°

AI CLUSTERED EVENT · 10/8/2026

UniPat Releases PaperBenchX: GPT-6 Astra Achieves Only 13.98% Reproduction Rate, Highlighting End-to-End Scientific Validation Gap

1 reports archived1 independent sourcesupdated 10/8/2026, 17:28:44
Synthesis & Latest Updates
1 Sources Cross-Validated

UniPat AI introduced PaperBenchX, a benchmark evaluating large models on 93 real-world paper reproduction tasks. Results show that even the top-performing GPT-6 Astra achieved only a 13.98% full reproduction rate, revealing a significant gap between solving specific math problems and reliable end-to-end scientific replication. The benchmark compiles scientific conclusions into executable, replayable, and verifiable workflows to establish new standards for measuring AI progress in science.

LATEST/UniPat AI introduced PaperBenchX, a benchmark evaluating large models on 93 real-world paper reproduction tasks. Results show that even the top-performing GPT-6 Astra achieved only a 13.98% full reproduction rate, revealing a significant gap between solving specific math problems and reliable end-to-end scientific replication. The benchmark compiles scientific conclusions into executable, replayable, and verifiable workflows to establish new standards for measuring AI progress in science.

TIMELINECoverage timeline

Total 1 reports · Latest first
  1. 量子位官网T2·78 pts

    UniPat Releases PaperBenchX: GPT-6 Astra Achieves Only 13.98% Reproduction Rate, Highlighting End-to-End Scientific Validation Gap

    Original: ChatGPT踢到铁板了！能破解千禧数学难题，但论文复现率低至13.98%？

    • GPT-6 Astra achieved only a 13.98% full reproduction rate across 93 paper reproduction tasks in PaperBenchX.
    • While AI can solve high-difficulty single-point problems like Millennium Prize challenges, it remains severely deficient in end-to-end replication of complex scientific workflows.
    • PaperBenchX converts paper conclusions into executable, replayable, and itemized verified scientific workflows, aiming to redefine evaluation standards for AI in science.