Learn
ChatGPT Work and Codex Limits by Plan: Free, Plus, Pro, Business
ChatGPT Work and Codex share one allowance. OpenAI estimates 15–160 GPT-6.1 Sol messages per five hours on Plus, Pro currently has no five-hour limit, and credits extend usage. See every plan, what uses limits faster and what to do at the limit.
ChatGPT Work and Codex draw on one shared allowance, and OpenAI publishes estimates rather than fixed limits. On ChatGPT Plus and a Standard Business seat, OpenAI estimates about 5–45 local messages every five hours on GPT-6 Astra, 15–160 on GPT-6.1 Sol, 15–150 on GPT-6 Sol and 350–3,000 on GPT-6 Luna, with weekly limits that may also apply. Pro plans currently have no five-hour limit. Free and Go get limited Work and Codex in the desktop app. When the allowance runs out, Plus and Pro subscribers can buy credits instead of waiting for the reset.
ChatGPT Work and Codex limits by plan
Plan | Price | Where Work and Codex run | Five-hour limit | What OpenAI publishes |
|---|---|---|---|---|
Free | $0 | Limited Work and Codex in the desktop app, on GPT-6 Luna (subject to rollout) | Not published | "Limited" access; no Codex Cloud |
Go | $8/month | Limited Work and Codex in the desktop app, on GPT-6 Luna (subject to rollout) | Not published | "Limited" access; no Codex Cloud |
Plus | $20/month | Work on desktop, web and mobile; Codex in the app, CLI, IDE, web and iOS, plus Codex Cloud | Yes | Per-model estimates (next table); weekly limits may apply; credits after the allowance |
Pro 100, 200 and 500 | $100, $200 or $500/month | Everything in Plus, with longer sessions | None currently | Pro 200 includes more usage than Pro 100, and Pro 500 the most; only Pro 500 has Ultrafast |
Business Standard seat | $25/month, or $20 billed annually | Work and Codex for each seat | Yes | The same per-model estimates as Plus; workspace credits after the allowance |
Business Premium seat | $125/month, or $100 billed annually | Work and Codex for each seat | None | Five times the Standard allowance |
Enterprise and Edu | Custom | Work and Codex as the workspace allows | Depends on pricing | With flexible pricing, no fixed rate limits and usage scales with credits; without it, the same per-seat limits as Plus for most features |
Plan descriptions and prices come from OpenAI's ChatGPT pricing page, its Codex pricing page and its Pro tiers and Business help articles, checked October 11, 2026. OpenAI does not publish a usage figure for Free or Go, or a multiple of Plus for any Pro tier.
Five-hour estimates by model
Model | Local messages per five hours (Plus and Standard Business) |
|---|---|
GPT-6 Astra | 5–45 |
GPT-6.1 Sol | 15–160 |
GPT-6 Sol | 15–150 |
GPT-6 Luna | 350–3,000 |
Source: OpenAI's Codex pricing page, checked October 11, 2026. OpenAI says these are estimates, not fixed message limits, and that cloud tasks may use more of the allowance than local messages. The wide ranges are the point: a short script edit and a long agent run on a large repository are both one message, but they use very different amounts of the allowance.
What uses the allowance faster
Usage depends on the model, the size of the task, how much context the agent holds, reasoning, tools and caching, so prompt length alone is a poor guide. A few rules are fixed:
- Work and Codex share one pool. A long research task in ChatGPT Work uses the same allowance as coding in Codex. On Plus and Pro, ChatGPT for Excel, Word and PowerPoint draw on it too.
- Speed modes multiply usage. Fast mode uses the included allowance at 2.5 times the Standard rate. Ultrafast, available only on Pro 500 among the Pro plans, uses it at 8 times the rate for GPT-6 Astra and GPT-6.1 Sol.
- Images cost more. Image generations use the included limits about three to five times faster on average than similar turns without an image, depending on quality and size.
- Voice in the desktop app counts. Voice in Desktop uses the Codex budget at $0.05 per minute, and tasks started by voice use the same pool.
- Some things are free. Auto-review safety checks do not count toward the limits when you sign in with ChatGPT.
ChatGPT's file-upload, image-creation and Chat voice limits are separate and do not apply to Codex. A banner such as an upload limit in ChatGPT does not mean your Codex allowance is used up.
How resets work
Plus and Standard Business seats have a five-hour window, and OpenAI says weekly limits may also apply. Your usage dashboard shows the current limits and reset times; in an active Codex CLI session, type /status. If you reach the limit in the middle of a turn, the agent can finish that turn, subject to fair-use limits.
OpenAI Support does not reset limits. OpenAI occasionally grants one-time resets, and some Plus and Pro referral promotions give a banked reset that refreshes both the five-hour and weekly windows when you use it.
What to do when you hit the limit
- Check which allowance ran out. The dashboard shows whether it is the five-hour or the weekly limit, and when it resets.
- Switch to a lighter model. GPT-6 Luna's estimate is roughly 20 times GPT-6.1 Sol's, so routine edits stretch much further on Luna.
- Buy credits on Plus or Pro. Credits continue Work and Codex after the included allowance, stay valid for 12 months and are not refundable. They are not API credits. Purchase amounts and prices are shown at checkout and vary by account and region.
- Use an API key for extra local work. Anyone can run additional local Codex chats with an API key, billed at standard API rates; see Codex credits vs API billing.
- Upgrade if it happens every week. Pro currently has no five-hour limit, and a Business Premium seat gets five times the Standard allowance with no five-hour limit.
Which plan fits
- Free or Go to try Work and Codex in the desktop app. Neither includes Codex Cloud.
- Plus for what OpenAI calls "a few focused coding sessions each week", or occasional Work tasks.
- Pro when the five-hour window stops you during the day. Choose the tier by how often you still hit the weekly limit; ChatGPT Pro 100 vs 200 vs 500 compares them.
- Business when the work belongs to a company workspace; give Premium seats only to the heaviest users.
- Enterprise or Edu with flexible pricing when you want usage to scale with credits instead of fixed limits.
For what Work is and how it differs from Chat, see What is ChatGPT Work?. For credit rates by model and the API-key route, see Codex pricing.
Evidence boundary
Official sources
Editorial guidance grounded in official product sources.
FAQ
Common questions
How many ChatGPT Work and Codex messages do I get on Plus?
OpenAI publishes estimates, not a fixed number. On Plus it estimates about 5–45 local messages every five hours on GPT-6 Astra, 15–160 on GPT-6.1 Sol, 15–150 on GPT-6 Sol and 350–3,000 on GPT-6 Luna. Weekly limits may also apply, and cloud tasks can use more than local messages.
Do ChatGPT Work and Codex have separate limits?
No. Work and Codex share one usage allowance and one credit balance, so a long Work task reduces what is left for Codex. On Plus and Pro, ChatGPT for Excel, Word and PowerPoint use the same allowance.
Does ChatGPT Pro have a five-hour limit for Work and Codex?
OpenAI says Pro plans currently have no five-hour limit, although weekly limits may apply. Pro 200 includes more usage than Pro 100, Pro 500 includes the most, and only Pro 500 offers Ultrafast, which uses the allowance at eight times the Standard rate.
What happens when I reach my Work or Codex limit?
The agent can finish the turn it is working on. After that, check the usage dashboard for the reset time. On Plus and Pro you can buy credits to keep going, switch to a lighter model such as GPT-6 Luna, use an API key for local work at API rates, or upgrade.
Do ChatGPT upload, image and voice limits count against Codex?
No. ChatGPT file-upload, image-creation and Chat voice limits are separate and do not apply to Codex. Image generation inside Work or Codex, and Voice in the desktop app, do use the shared Work and Codex allowance.
Do ChatGPT Business seats get more Work and Codex usage than Plus?
A Standard Business seat has the same five-hour estimates as Plus. A Premium seat, at $125 a month or $100 billed annually, gets five times the Standard allowance with no five-hour limit, and the workspace can buy shared credits for usage beyond the included limits.
Next steps
Take the next buying step
Use these next pages to confirm the plan, tool, or alternate route that fits once the spend boundary is clear.