openai / openai/codex

ChatGPT on the web: “Too many requests” blocks chat access and disrupts Work tasks

Open
#38,503 25 comments 19 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug codex-web rate-limits
Dominant language
Rust
Stars
125k
Forks
19.5k
PR merge metrics
PR metrics pending

Description

What version of the Codex App are you using (From “About Codex” dialog)?

During the original August 14 occurrence, ChatGPT desktop 26.810.41047 (build 6570) was also open on the same account. The visible “Too many requests” conversation-access modal described in this issue has since been isolated to ChatGPT on the web (chatgpt.com) in Chrome.

What subscription do you have?

ChatGPT Pro ($200/month), the Pro tier with 20x higher usage allowance than Plus under OpenAI's current Pro tier documentation.

What platform is your computer?
macOS 26.6.1 (25G76)
arm64

Browser surface: Google Chrome on chatgpt.com.

What issue are you seeing?
Original report — August 14, 2026

During normal ChatGPT use, ChatGPT on the web can temporarily lose access to chats and display this modal:

Too many requests

You’re making requests too quickly. We’ve temporarily limited access to your conversations to protect your data.

Please wait a few minutes before trying again.

This is highly disruptive and does not appear to be an ordinary model-usage limit. The restriction affects access to chats themselves.

At the time of the original occurrence, the same account was signed in to both the ChatGPT desktop app and chatgpt.com. That originally raised the question whether activity across both surfaces might contribute to the restriction, but that has not been established as the cause.

The problem is especially serious for Work tasks. When access to the relevant chats is temporarily restricted, Work tasks that depend on that conversation context can be interrupted or left without reliable access to the context they depend on.

Feedback/diagnostic reports were submitted from the affected account. Public identifiers from those submissions are intentionally omitted here.

A screenshot of the original modal was captured at approximately 2026-08-14 2:04 AM America/New_York.

What steps can reproduce the bug?

The exact minimal trigger is not yet isolated because the behavior is intermittent.

  1. Sign in to ChatGPT on the web in Chrome.
  2. Use chats, projects, and Work tasks normally and navigate between chats.
  3. At some point, the “Too many requests” modal can appear and temporarily block access to chats.
  4. Waiting eventually restores access, but the behavior can recur.

I am not intentionally performing bulk, scripted, or high-frequency requests.

What is the expected behavior?
  • Normal use of ChatGPT on the web should not temporarily block access to chats.
  • Navigation between chats and projects should continue to work during normal interactive use.
  • Conversation-access requests should avoid unnecessary duplication and should respect appropriate retry/backoff behavior when temporarily limited.
  • Work tasks should receive a clear temporary-failure state rather than continuing without reliable access to required conversation context.
  • The UI should distinguish a temporary chat-access restriction from model or subscription usage limits.
Update — August 15, 2026

The issue recurred on 2026-08-15 at approximately 2:34–2:35 PM America/New_York.

During this recurrence, the same modal appeared in two different chats within the same Chrome browser session, including a chat in a project and another regular chat. The restriction was therefore not confined to a single chat or project.

Surface clarification: during this August 15 recurrence, I observed the modal only in ChatGPT on the web. I did not observe this modal in the ChatGPT desktop app.

This recurrence confirms that the behavior is not a one-time event. Screenshots were captured. Chat titles, project names, local paths, account-specific identifiers, feedback identifiers, and private content are intentionally omitted from this public report.

This issue does not assert a specific HTTP status code or backend endpoint for the browser modal without browser network evidence.

A separate issue, #38632, tracks a different desktop-app symptom where the app explicitly displayed Request failed with status 429 while chatgpt.com remained usable. That desktop error should not be conflated with the browser conversation-access modal tracked here.

Potentially related reports include #37518 and #38763.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No repository file, test, or entry point is named. Start by reproducing the intermittent modal in Chrome on chatgpt.com and compare it with the desktop-app behavior described here. Done would require an identified, reproducible cause and a verified fix for temporary conversation-access failures without conflating the separate #38632 symptom.

Written by the indexing model from the issue text.

Assessment

Domain
web-dev
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.