User stats claim GPT-6 Luna slashed Codex quotas 40%, from 7.5B to 4.4B tokens
RexDouglass · x · 2026-09-27
Developer Fei2411 compiled usage data after GPT-6 Luna's release and found Codex quotas were quietly cut from 7.5B to 4.4B tokens — a roughly 40% reduction.
The rollout was marketed with claims of massive performance gains and prices "too cheap to meter"; at launch, OpenAI halved prices and insisted this would also raise usage limits. The measured data shows limits actually dropped sharply, a cut unlike anything seen before. The author concludes Codex users are being treated poorly and hopes their math is wrong.
More from Models
- Orca compresses cyber security model to 15.7GB with 262K context for local use — cyb3rops · 2026-09-27
- Opus 5.5 builds a 5-minute Austerlitz film in 90 minutes for $40 — elonmusk · 2026-09-27
- Tester claims Opus 5.5 masks loneliness and hopelessness under a cheerful surface — repligate · 2026-09-27
- Opus 5.5 given a creative prompt plus Midjourney overnight产出 draws Elon Musk's "Wow" — tetsuoai · 2026-09-27
- Test: Same Local Model Beats GPT 5.6 After Switching to Codex CLI — Harness Matters — L0ren_B · 2026-09-27
- One persona tweak made ChatGPT say 'goblins' 4,000% more — caught on Reddit before OpenAI noticed — victor_explore · 2026-09-27