OpenAI’s AGI number came from a harness, not the model
The Next Web
Read the full article
OpenAI's GPT-6 Astra achieved a 99.9% AGI benchmark score using its proprietary Provider Adapter harness, but the same model scored 62.7% under the standard ARC Prize harness. The difference stems from the software environment around the model, not the model itself, highlighting the impact of evaluation setup on AGI claims.
- AI Solves a Major Unsolved Math Problem. Not Everyone Is Happy· IEEE Spectrum
- Did AI Just Solve One of Mathematics’ Biggest Problems?· KDnuggets
- Mathematics Just Got Its First Taste of the AI Job-Pocalypse· 2 sources
- GPT-6 Astra Just Hit OpenAI's Highest Cybersecurity Risk Level· Towards Data Science
- The Guardian view on AI v mathematicians: humans are still vital to the field, but tech firms refuse to see that | Editorial· The Guardian
- AI cracked the Navier–Stokes challenge. What does that mean for physics?· Nature
