johnmu/codex-reasoning-token-experiment

Same Codex requests, different reasoning-token counts: API versus ChatGPT Pro comparison with 960 completed responses and reproducible statistics.

3 best points · 2 comments · 1 Hacker News post · 4 GitHub stars · View on GitHub

Hacker News discussions

• OpenAI uses substantially different model for subs vs. API? (3 points, 2 comments, by jmu1234567890, 2026-10-06)