GPT-6 Astra tops Mercor's APEX-Agents leaderboard at 62.4% on professional tasks

sherwinwu · x · 2026-09-04

GPT-6 Astra debuts at #1 on Mercor's APEX-Agents leaderboard with a 62.4% mean score and 46.7% Pass@1, edging Claude Fable 5.1 by 0.4 points on mean. It gains 5.7 points on mean and 6.7 on Pass@1 over GPT-5.6 Sol, improving all three professional domains (banking, consulting, law) by 4.5+ points. The benchmark spans 33 simulated work worlds and 480 tasks; the dataset and the Archipelago eval infra are open-source.

Original post →

More from Models

Models channel →