Codex Windows: 5-hour limit exhausted in a few minutes / resets used up #52150

Activity

1. github-actions

   added

   bugSomething isn't workingSomething isn't working

   appIssues related to the Codex desktop appIssues related to the Codex desktop app

   windows-osIssues related to Codex on Windows systemsIssues related to Codex on Windows systems

   rate-limitsIssues related to rate limits, quotas, and token usage reportingIssues related to rate limits, quotas, and token usage reporting

   on Oct 8, 2026
2. github-actions

   changed the title [-]Codex Windows: limite 5 ore si esaurisce in pochi minuti / reset consumati[/-] [+]Codex Windows: 5-hour limit exhausted in a few minutes / resets used up[/+] on Oct 8, 2026
3. github-actions commented on Oct 8, 2026

   

   github-actionsbot

   on Oct 8, 2026 – with GitHub Actions

   Contributor

   More actions

   English translation:

   Component: Desktop app (Windows)
   Steps to reproduce:

   1. Open a local project in the Codex Desktop app.
   2. Work in a long conversation using Stata do-files and local data files.
   3. Use the model configured in the app, with repeated code execution, log inspection, and file analysis.
   4. After several minutes of work, check Settings → Usage.
   5. Continue working after the quota resets and repeat the same workflow.
      Expected:
      The 5-hour usage quota should not be exhausted unusually quickly when using a normal coding workflow. Repeated executions of local commands and inspection of moderate-sized outputs should not cause the quota to reset multiple times in the same day without a clear indication of what consumed it.
      Actual:
      The 5-hour Codex quota reached 0% remaining twice during the same day. The account showed the 5-hour limit at 100% used, while the weekly limit was only about 16% used. After the limit was reached, the app displayed a message indicating that the usage limit had been reached and suggested switching to a faster/lighter model. No additional credits were available.
      Evidence:
      Screenshots of:

   

   Session ID:no-active-thread-01a11c0c-29ce-7a70-994f-b4fe8a9200fd
   Additional context:
   The workflow included local Stata do-files, inspection of logs, analysis of CSV/DTA files, and repeated iterations in a long conversation. Please clarify whether local command execution, long conversation context, tool calls, or repeated file analysis can consume the same 5-hour Codex quota, and provide a per-session or per-operation usage breakdown if possible.
4. github-actions commented on Oct 8, 2026

   

   github-actionsbot

   on Oct 8, 2026 – with GitHub Actions

   Contributor

   More actions

   Potential duplicates detected. Please review them and close your issue if it is a duplicate.

   • Resetting the usage limit and getting it drained only in 5 minutes #51445
   • Sharp reduction in usage #51744
   • (5-hour) token usage window decrease? #52105
   • [Usage limits] GPT-6 Astra Low/Standard: 5h allowance exhausted after ~20–30 min of resumed use (Oct 8) #51951
   • [macOS][Codex Desktop] Sudden ~60% increase in 5h quota usage overnight; large prompt-cache miss on small turn #51604

   Powered by Codex Action
5. stmoeller62 commented on Oct 10, 2026

   

   stmoeller62

   on Oct 10, 2026

   More actions

   ChatGPT Plus – Severe reduction in usable Codex time

   I am experiencing a substantial and persistent reduction in usable Codex time.

   In June/July 2026, I was able to work with Codex for several hours on complex Windows development and system maintenance tasks without encountering significant usage limitations.

   Currently, my available usage is sometimes exhausted after approximately 20–30 minutes.

   Switching to less demanding models and lower reasoning settings has not significantly improved the situation.

   I have contacted OpenAI Support several times, but have not received a satisfactory explanation. My latest inquiry may not have received any response.

   I understand that usage limits exist and that complex tasks consume more resources. However, the dramatic difference compared with earlier months raises concerns about possible changes or errors in usage accounting.

   Could the Codex team please investigate:

   • Whether recent changes have increased quota consumption for comparable workloads.
   • Whether account-specific rate limiting or incorrect usage accounting could be involved.
   • Whether excessive context processing or unexpected cache invalidation could explain the behavior.

   This issue significantly affects my ability to use Codex for serious software development.

   I would appreciate a technical explanation rather than another general description of subscription limits.
6. raviknits commented on Oct 11, 2026

   

   raviknits

   on Oct 11, 2026

   More actions

   Repeated runs plus log inspection in one long conversation is probably most of it. Each Stata log the agent reads stays in the conversation and is sent again with every later turn, so after a dozen runs, every message carries all of the old logs. Starting a new thread per analysis step, or asking Codex to print only the tail or the errors of a log, cuts it a lot.

   If you want to check, there's an offline script on densilo.com/tokens that reads your local Codex sessions and shows how much of what was sent was command output