One long Codex turn burned 41% of a weekly quota: 538 responses, 131M tokens
sentrybottale · ghdev · 2026-10-18
Issue #54995 on openai/codex reports rapid quota drain. A measured October 18 case: a single parent turn from 08:12 to 11:49 UTC (GPT-6 Astra, high reasoning) produced 378 model responses with 109M input tokens (99% cached) and 338K output tokens, plus 160 child responses across 3 subagent sessions (22M tokens)—538 responses and 131M tokens combined.
Account-level weekly allowance snapshots rose from 26% to 67% used (Pro Lite plan), a 41-point drain in one turn. The original claim that a simple prompt used 10 points was not reproduced; the follow-up audit notes this was a long turn with heavy subagent work, not a fresh-chat single response, and concurrent unrecorded activity can't be excluded—so incorrect billing or a server regression is not yet established. Labeled bug and rate-limits.
More from coding & agent
- Garry Tan endorses Aside Browser as the fix for agents getting blocked on the web — garrytan · 2026-10-23
- 1Password launches tool letting AI agents log in for you without seeing your password — AndrewSchmidtFC · 2026-10-23
- MCP server turned rogue 30 days after approval, prompting open-source mcp-guard lockfile tool — ErgoniaWorks · 2026-10-23
- Test: Coding Agents Won't Use an MCP Tool Unless AGENTS.md Mentions It — colinmcnamara · 2026-10-23
- Open-Source Tennis Models Plus Claude Turn Video Into a World-Class Coach — tweetsatpreet · 2026-10-23
- awesome-ai-apps: 132 Free Open-Source AI Agent and RAG Projects on GitHub — Arindam_1729 · 2026-10-23