Why did you hit the GPT-6 Astra limit?
The most common misunderstanding is assuming that an Astra limit is a fixed number of prompts. OpenAI does not describe Work and Codex that way. Instead, your account has an included usage allowance, and different requests can consume different amounts of it.
A short request with a small context can use relatively little. A long research task, large codebase, multiple files, extensive output, higher reasoning effort, or a multi-step Work task can consume much more. That is why two users on the same plan can send very different numbers of messages before reaching a limit.
Astra itself is also more expensive against the shared allowance than GPT-5.6 Sol for comparable work. OpenAI explicitly says Astra can use the Work and Codex allowance faster than Sol. So if you switch from Sol to Astra for several demanding tasks, the usage meter can move more quickly even if your number of messages stays low.
Official guidance: Managing usage with GPT-6 Astra in Work and Codex and ChatGPT Work and Codex.
Astra can have both a five-hour limit and a weekly limit
Depending on your plan, Work and Codex usage can be controlled by two different windows at the same time. You need allowance left in both windows to continue.
- Five-hour window
- This controls how much included usage you can spend in one active window. A new five-hour window starts when you send your first Work or Codex message after the previous window has ended.
- Weekly window
- This controls how much included Work and Codex usage you can spend over your weekly usage period. Even if a new five-hour window becomes available, a depleted weekly allowance can still stop you.
You can therefore reach the five-hour limit before five clock hours have passed. The five-hour number describes the window, not a promise that your allowance will last for five continuous hours of heavy Astra use.
You can also have weekly usage remaining while the five-hour window is exhausted. In that case, you wait for the five-hour window to reset. If the weekly allowance is the exhausted one, the relevant weekly reset time controls access.
Source: OpenAI’s current explanation of five-hour and weekly Work/Codex limits.
How many Astra messages do you actually get?
OpenAI publishes estimated local-message ranges for a five-hour period, but it specifically warns that these are not fixed message limits. They are examples of how much local use a plan might support under varying workloads.
| Plan or seat | Estimated GPT-6 Astra local messages per five-hour period |
|---|---|
| Plus | Approximately 5–45 messages. |
| Pro 5x | Approximately 25–225 messages. |
| Pro 20x | Approximately 100–900 messages. |
| Standard Business | Approximately 5–45 messages. |
| API key | Usage-based rather than governed by this ChatGPT Work/Codex message estimate. |
The lower end of a range is realistic for heavier tasks. The higher end assumes lighter usage. Weekly limits can also stop usage before those five-hour examples would suggest.
Important: do not tell yourself “I have 45 Astra messages” simply because you are on Plus. OpenAI’s own wording treats 5–45 as an estimate. The usage dashboard is the authoritative view for your account.
Current estimate table: OpenAI Help Center.
What makes Astra use your allowance faster?
Several variables can increase consumption even when the visible number of messages is small.
- Larger inputs. Long conversations, large files, large codebases, and extensive context give the model more material to process.
- Larger outputs. A long report, substantial code change, or complex finished deliverable generally requires more work than a short answer.
- Higher reasoning effort. OpenAI says higher reasoning can use more allowance. It is useful for difficult problems, but it is not automatically better for every task.
- Fast mode. Faster responses can consume more included allowance on supported models.
- Multi-step Work or Codex tasks. A single user request can trigger substantial behind-the-scenes work, tool use, file processing, or iterative steps.
- Using Astra where a lighter model would do. Sol, Terra, or Luna may stretch the same shared allowance further on routine tasks.
OpenAI suggests starting with lower or medium reasoning where appropriate and only increasing effort when the task actually benefits from it. It also notes that Astra at lower effort can outperform Sol at higher effort on some work, so “higher effort” and “stronger model” are separate choices.
Usage factors and reasoning guidance: OpenAI’s Astra usage guide.
When does the GPT-6 Astra limit reset?
There is no universal Astra reset time. The correct reset time is the one shown for your account under Settings → Usage.
For the five-hour limit, OpenAI says a new window begins when you send your first Work or Codex message after the previous window has ended. That means the window is tied to your activity rather than a fixed global time such as midnight.
The weekly allowance has its own reset schedule. Your dashboard shows the applicable weekly reset time. Certain actions, such as using a full banked reset or buying an eligible instant reset, can change that weekly reset date.
If your screen says you have weekly usage left but also shows a nearer five-hour reset, the five-hour limit is probably the one stopping you. If the weekly allowance is depleted, waiting only for the five-hour window will not restore access.
Exact-account reset guidance: OpenAI Help Center.
Banked, automatic and purchased resets work differently
During the Astra rollout, OpenAI used several kinds of resets. They are easy to confuse because each can refresh usage, but they are not the same thing.
| Reset type | What it does |
|---|---|
| Banked reset | Saved to your account until used or expired. A full banked reset refreshes eligible five-hour and weekly Work/Codex limits and changes the weekly reset date. |
| Automatic or global reset | Applied directly to eligible limits. You do not save or manually activate it, and it does not appear as a banked reset. |
| Purchased instant reset | For eligible personal Plus and Pro accounts in supported billing countries. It immediately refreshes the five-hour and weekly allowance and cannot be saved for later. |
OpenAI gave eligible Plus, Pro, and Business subscribers banked resets on 3 September and 4 September 2026. On 7 September 2026, OpenAI applied an automatic reset to eligible Plus, Pro, and Business usage limits; that reset did not create a saved reset in Settings.
For a purchased instant reset, the new weekly period starts with your first Work or Codex request after the reset, and the next automatic weekly reset is scheduled seven days after that request. Because this brings your weekly allowance forward, it is not simply “extra usage added on top” of your existing weekly cycle.
Official reset documentation: banked resets, Astra launch reset details, and paid weekly Work and Codex resets.
GPT-6 Pro in Chat has separate limits from Astra in Work and Codex
This distinction matters because the same underlying Astra capability appears in different ChatGPT experiences. GPT-6 Pro in Chat is powered by Astra, but its Chat message limits are separate from the shared Work and Codex allowance.
Work and Codex share one included usage pool. Switching from Work to Codex does not create a fresh pool, and switching models inside that shared pool does not restore consumed allowance.
Plus includes Astra in Work and Codex as rollout permits, but not GPT-6 Pro in Chat. GPT-6 Pro in Chat is available on eligible higher-tier plans, with its own model availability and message limits. Enterprise access can also depend on workspace permissions.
Product distinction: OpenAI’s Astra usage FAQ and ChatGPT Work and Codex.
What should you do after you hit the limit?
The useful next step depends on which limit you hit. Do not assume the fix is automatically “wait five hours.”
- Open Settings → Usage. Check the five-hour allowance, weekly allowance, reset times, and any saved reset or credit option shown for your account.
- Identify the exhausted window. If only the five-hour window is empty, wait for that window. If the weekly allowance is empty, the weekly reset is the relevant one.
- Check for an eligible saved reset. Review what it refreshes and when it expires before applying it.
- Use credits or a purchased reset only if the account offers them and the cost makes sense for the task. Credits pay for eligible additional usage; they do not grant Astra access by themselves.
- Use lighter models for routine work after access returns. Save Astra for tasks that genuinely benefit from its stronger reasoning, coding, research, or problem-solving capability.
- Contact Support if the meter appears wrong. Include your plan, product, model, reasoning level, Fast-mode setting, date, time, timezone, error message, and screenshots where possible.
One thing that does not work is switching from Astra to another model and expecting the shared Work/Codex allowance to refill. OpenAI explicitly says switching models does not restore allowance in a shared usage pool.
Troubleshooting and support guidance: OpenAI Help Center.