GPT-6.2 Astra Tops Agents' Last Exam Benchmark

OpenAI's GPT-6.2 Astra set a new record on Agents' Last Exam v1.1, beating GPT-6 Astra by 5.2 points and Claude Fable 5.2 by 5.5 points, while cutting per-task cost to under a third (about 5% by one estimate).

2026-11-18 ~ 2026-11-18 · 2 related posts