Skip to content
Brand AI

GPT-6 Astra

Recent events

Recent events tied to this brand in public sources.

Back to profile1 entries

  1. 2026

    Claude Fable 5.1 leads OSWorld-Science scientific AI agent benchmark evaluation

    OSWorld-Science, a computer-use agent benchmark for real scientific software scenarios developed by a Tsinghua University-led research team with multiple universities and research institutions, comprises 146 expert-designed tasks across 7 scientific fields and 14 scientific software configurations and was used to evaluate 12 vision-language models. The 12 models averaged a 34.3% task score; Claude Fable 5.1 led with 73.7%, ahead of Claude Opus 5 (57.9%), GPT-6 Astra (57.4%) and GPT-5.6 sol (50.1%), while Gemini 3.1 Pro scored 17.2%.

    Source