Codex runs GPT-6 with far less reasoning than the API — up to 12x lower on max effort
RexDouglass · x · 2026-10-07
Developer chetaslua measured the same GPT-6 model inside Codex and found its reasoning "juice" is lower than the API at every effort level — up to 12x less on max: model Astra shows 832 juice via API vs 128 in Codex, and Sol shows 768 vs 62.
A quoted reply from Anthropic's joannejang jokes "did they quantize my boy astra," suggesting OpenAI may have quantized the model or capped thinking budgets inside Codex to cut inference costs.
More from coding & agent
- Developer has Codex build a Solarman solar-monitoring macOS app in 3 minutes — Dimillian · 2026-10-07
- Open-source Claude skill huashu-art-motion: AI-made science video hits 100K views on Douyin in 5 hours — AlchainHust · 2026-10-07
- Redditor passes three Claude certification exams and shares his full study playbook — Few-Association-913 · 2026-10-07
- universal-modder: open-source AI agent mods any game and leaves pitfall notes for the next agent — sujingshen · 2026-10-07
- Slides released: 'Ultimate guide to multi-harness RL' talk from Kernel Panic Madrid — SergioPaniego · 2026-10-07
- AI agents' next hurdle: websites like Amazon are locking them out — emmanuelvivier · 2026-10-07