ChatGPT “Too Many Requests in 1 Hour” — What It Means and How to Fix It
“You’ve reached our limit of messages per hour. Please try again later.” is ChatGPT’s consumer message cap, not a bug and not the developer API’s HTTP 429. This guide explains why it triggers faster than you expect, how the reset actually works, and what to do right now.
What does “too many requests in 1 hour” mean on ChatGPT?
This message is a per-account, per-model quota on the ChatGPT consumer product (web and app), not an outage and not the same system developers hit when calling the API:
| Surface | What causes it | Fix |
|---|---|---|
| ChatGPT Free | Small GPT-5/GPT-4o-class quota, then auto-downgrades to a mini model | New chat, switch model, or wait |
| ChatGPT Plus / Pro | Much higher per-model rolling-window quota | Same fixes; upgrade raises it further |
| “Too many requests in 1 hour” | The consumer message cap on chat.openai.com / chatgpt.com | Follow the 5 steps below |
| HTTP 429 (API) | Your OpenAI API account’s RPM/TPM/quota limit — a separate developer-side system | See openai-429-error guide |
The key distinction: the consumer message cap governs how many turns you can send in the ChatGPT app inside a rolling window of roughly the last 1–3 hours. The API’s 429 governs how many requests-per-minute or tokens-per-minute your developer account can send to platform.openai.com. They use different infrastructure, different dashboards, and fixing one does nothing for the other.
5 steps to fix and avoid the message limit
Rule out an OpenAI incident first
Before assuming this is your personal cap, confirm OpenAI isn’t degraded. It’s rare, but during real incidents the error text can be misleading — the system may be throttling everyone, not just you.
Diagnosis checklist
- Open prismix.dev/service/openai — active incidents appear at the top.
- If there’s no incident and the error clears once you open a brand-new chat, it was your personal quota — go to step 2.
- Also check status.openai.com directly for OpenAI’s own incident page.
Start a new, shorter chat (the biggest lever)
This is the single most effective fix and the one most people miss. ChatGPT re-sends the entire conversation history as context on every turn — a 40-message thread costs vastly more than 40 fresh single-turn chats, because the model has to reprocess the whole growing thread each time.
Why long chats drain the cap so fast
- Turn 1 sends a short prompt; turn 50 in the same thread sends that prompt plus 49 prior exchanges as context, every time.
- Switching topics inside one long-running chat is the most common way people burn through the hourly cap without noticing.
- If you’re debugging code or working through a document, copy the key context into a fresh chat instead of scrolling one thread for hours.
Switch to a lighter model in the model picker
Message caps are tracked separately per model. If your flagship reasoning model is capped, a mini/instant-class model in the picker often still has quota available — and generally carries a more generous cap to begin with.
- Open the model picker at the top of the chat window and choose the lighter/instant option instead of the top-tier reasoning model.
- Free-tier accounts are auto-switched to a mini model once the higher-tier quota is used up — that’s expected behavior, not an error.
- For tasks that don’t need deep reasoning (rewriting, quick lookups, formatting), defaulting to the lighter model avoids hitting the cap at all.
Understand the rolling window before you wait a full hour
“Per hour” is approximate. The cap is enforced against a rolling window of roughly your last 1–3 hours of messages, not a fixed clock hour that resets on the dot. As your oldest messages inside that window age out, room opens up gradually.
Practical takeaway
- Try again after 10–20 minutes rather than assuming you must wait a full 60.
- If you were blocked mid-burst (many messages sent quickly), the window clears faster than if you were steadily chatting for hours.
- There is no visible countdown timer in the ChatGPT UI — treat the “1 hour” wording as a upper-bound estimate, not a guarantee.
Upgrade tier — or use the API if this is really developer traffic
If you consistently hit the cap during normal use, ChatGPT Plus or Pro raises the message quota substantially over Free. Upgrading does not remove the cap, but it makes it far less likely to matter for typical daily use.
If instead you’re scripting or automating requests and seeing this error (or an HTTP 429), you’re very likely calling ChatGPT through browser automation against the consumer product rather than the API — switch to the official OpenAI API, which is billed and rate-limited separately from ChatGPT:
# Python — the API's rate limit is separate from the ChatGPT app's message cap
import openai
try:
response = client.chat.completions.create(
model="gpt-5",
messages=[{"role": "user", "content": "Hello"}],
)
except openai.RateLimitError as e:
print(f"API rate limit hit: {e}") # not the same as the ChatGPT app's hourly capFor that developer-side 429, see /guides/openai-429-error and /guides/openai-rate-limit-exceeded.
Get an email the next time OpenAI goes down
Outage alerts for OpenAI, straight to your inbox. No account needed, unsubscribe in every email.
Watching more than one service? A free account covers 5 services + a daily digest — and Pro is currently free.
FAQ
How long until the ChatGPT limit resets?
The cap uses a rolling window of roughly the last 1–3 hours, not a fixed clock hour, so there’s no single countdown shown anywhere in the app. As older messages inside that window age out, capacity frees up gradually — you can often send again within 10–20 minutes even though the error text says “per hour.”
Does the limit reset exactly after 1 hour?
No. “Try again in an hour” is an approximation, not a hard reset timer. Because the cap is based on a rolling window of your recent messages, it can clear before or after the 60-minute mark depending on exactly when your earlier messages in that window were sent.
Is this the same as API error 429?
No. This message is specific to the ChatGPT.com consumer app (Free, Plus, Pro) and reflects your personal message quota. HTTP 429 is a developer-facing error from the OpenAI API tied to your account’s requests-per-minute or tokens-per-minute limits. They’re unrelated systems — see openai-429-error.
Does ChatGPT Plus remove the message limit?
No. Plus and Pro raise the cap substantially over Free, and each model tracks its own separate quota, but no paid tier removes limits entirely. Very long conversations or heavy daily use can still hit the cap even on Pro.
Why did I hit the limit so fast?
Almost always a long conversation. ChatGPT resends the full thread as context on every turn, so one long-running chat burns quota far faster than the same number of messages spread across fresh chats. Large file/image uploads and higher-effort reasoning models also count more heavily.