Unverified Agents' Last Exam v1 chart: Claude Opus 5.5 tops score, GPT-6.1 Terra best value
SEDIDEL · x · 2026-10-27
Another unverified OpenAI-style leaderboard chart, labeled Agents' Last Exam v1, shows Claude Opus 5.5 leading at 63% score but costing over $9; GPT-6.1 Terra achieves 57-59% at under $1, the best value; GPT-6 Astra matches 58% at roughly $8, while GPT-6 Luna ranks weakest.
Like the Terminal-Bench 4.0 chart, it's a rumor-grade leak pending official confirmation.