Mercor Claims GPT-6 Astra Leads APEX-Agents Leaderboard
OpenAI researcher retweets Mercor post announcing new AI agent leaderboard results.
TLDR
Ted Sanders, research engineer at OpenAI, retweeted a post from Mercor stating that GPT-6 Astra debuted at number one on the APEX-Agents leaderboard. The post lists a 62.4% mean score in first place and 46.7% Pass@1 in second place, ahead of Fable 5.1. It points to the Mercor site covering benchmarks for AI agents on long-horizon, multi-tool professional tasks. The evidence contains only this social media claim with no further confirmation shown.
Combined views
32.2K
3 Sources, first seen 27d ago