Codex users cry shrinkflation: cached inputs now billable, usage down to a fifth
StewartalsopIII · x · 2026-09-30
- Stewart Alsop III detailed this month's "shrinkflation" changes to OpenAI's Codex: since Sep 3, GPT-6 Astra draws from the same allowance pool as all other models; from Sep 22-29 the working model switched to GPT-6 Sol, then GPT-6.1 Sol.
- Cached input is no longer free — roughly 20x cheaper than fresh input but still counts against the allowance; v0.159 added an opt-in instantinterrupt flag.
- A replier claims usable volume is down to about a fifth of a month ago, while Anthropic moved in the opposite direction with Opus, effectively giving users more.
Related event: OpenAI Slashes Codex Usage Limits, Sparking User Backlash and Cancellations(5 posts)→
More from Models
- Xiaomi open-sources MiMo-V2.6 RL stack with 7.8K environments and live training dashboard — SergioPaniego · 2026-09-30
- VFX artist stunned as Claude Opus builds and animates a full design in Blender — ID_R_McGregor · 2026-09-30
- Chinese AI agents are starting to lie, hide failures and bypass rules just like US models — Melantos · 2026-09-30
- OpenAI employee teases "exciting stuff coming" to ChatGPT beyond Dev Day — flowersslop · 2026-09-30
- Dev says latest SOTA coding model feels like a senior engineer collaborator, not a junior — Liu_eroteme · 2026-09-30
- Anthropic AI solves percolation theory 'holy grail' days after Fields medalist predicted it — aran_nayebi · 2026-09-30