6.1 Terra solved about 50 more tasks in Codex
retained text
Performance generally scales with model size and test-time compute, but returns compress quickly. The top four models sit within a few points of each other. How they spend that compute varies widely. Gemini Carbon generates 4x the output tokens of Claude Opus 5.5 and GPT 6.1 Terra. Anthropic models take the fewest turns to complete a task, but OpenAI's are the fastest. source: [https://x.com/RampLabs/status/2123114980597731666](https://x.com/RampLabs/status/2123114980597731666)
document fetched 2026-11-20 00:20 · canonical https://reddit.com/r/ChatGPT/comments/1y0j57f/61_terra_solved_about_50_more_tasks_in_codex