Rogo evaluation: GPT-6.2 Astra financial accuracy 61%, surpassing Opus 5.5
OpenAI Launches GPT-6.2: What Our Evaluations Reveal | Rogo
AI summary
OpenAI releases GPT-6.2 Astra and Sol; according to Rogo evaluation, Astra scores 61% on BigFinanceBench, surpassing Claude Opus 5.5's 53%. Sol scores 58%, on par with GPT-6.1 Sol, with 15% faster responses and 13% fewer input tokens at mid-tier reasoning. Presentations are still led by Claude Opus 5.5, with a 94.8% blind-evaluation preference rate.
Read original