Claude vs ChatGPT usage limits: who still has a 5-hour cap?

If you work in an AI agent all day, the usage limit matters more than the model name. Both Anthropic and OpenAI now describe their limits differently, and the plan you pick decides whether you hit a wall every few hours.

This comparison was checked against each company’s official pricing and help pages on November 3, 2026. Limits change at the companies’ discretion, so treat it as a dated snapshot.

The short answer

• Claude Ultra ($200 a month): no five-hour limit for verified accounts. There is still a weekly cap of 10 times Pro.
• ChatGPT Pro (from $100 a month): OpenAI’s Codex pricing page says Pro plans “currently have no five-hour limit.”
• Everyone else: Claude Free, Pro and Max, and ChatGPT Plus and Business, still use a rolling five-hour window, plus weekly limits.

Claude: plan by plan

Anthropic’s pricing page lists these usage terms:

A few details from Anthropic’s help pages:

• How the window works. The five-hour period starts when you send your first message, and your limit resets in full five hours later. Weekly limits reset at a fixed time each week assigned to your account.
• One shared pool. Claude on the web, desktop, mobile and Claude Code all count toward the same limit.
• Verified Ultra accounts. Anthropic says it automatically verifies Ultra accounts using factors such as account and payment history, to help prevent fraud and abuse. Until an account is verified, a temporary limit applies and is removed automatically. To check, open Settings, then Usage: if you do not see a “Current session” bar, your account is verified. Verifying your phone number at checkout is a separate step and does not remove the limit.
• Fable. Anthropic’s plan table lists a separate weekly limit for Fable. On Max and Ultra, Fable is included up to 50% of the weekly limit. On Pro it is paid for with usage credits.
• Other limits. Anthropic says it may also limit usage with weekly and monthly caps, or by model and feature, and does not guarantee uninterrupted service.
• Going past a limit. On paid plans you can switch on usage credits and keep working at standard API rates. Anthropic’s Ultra help page also says that if you have a limit reset, you can use it to restore your weekly usage to full.
• The old name. Ultra is the former Max 20x plan, renamed with the same price. Existing subscribers moved over automatically. The “20x” referred to the five-hour window, while the weekly allowance is 10 times Pro. We covered the change in Claude Max 20x becomes Ultra.

ChatGPT and Codex: plan by plan

OpenAI’s Codex pricing page says ChatGPT Work and Codex share usage, and that ChatGPT Work usage inside ChatGPT uses the same pricing, credits and limits as Codex.

OpenAI publishes estimates, not guarantees, for local Codex messages per five-hour period on Plus and standard Business:

OpenAI stresses that these are not fixed message limits. The number depends on the model, the task’s size and complexity, and whether it runs locally or in the cloud. Local messages and cloud chats share one allowance, and weekly limits may also apply. Fast mode uses 2.5 times the included usage, and Ultrafast mode on GPT-6.1 Sol uses 8 times, relative to Standard mode. Ultrafast access is listed for the $500 Pro plan.

OpenAI’s page also notes that its API prices are separate from subscription usage and should not be used to estimate how many tasks a plan includes.

What this means in practice

• Occasional intense sessions: the five-hour window is the one you hit. Claude Ultra for verified accounts and ChatGPT Pro are the plans without it.
• Steady all-day use: the weekly cap is the real limit. On Claude, Ultra’s is 10 times Pro. OpenAI says weekly limits may apply but does not publish the figure on the pricing page.
• Choosing by model: on ChatGPT, heavier models use much more of the window. Moving from GPT-6 Astra to GPT-6.1 Terra raises the estimated local messages per five hours from 5 to 45 up to 35 to 285.
• Team and compliance buyers: both vendors offer Team, Business or Enterprise plans with different terms. Check the plan page for your organization.

How this has changed

Limits have moved a lot this year. OpenAI briefly removed its five-hour limit in July, and Anthropic cut weekly usage by 17% for Claude Code users. OpenAI has since added an Ultrafast mode and more flexible ways to keep working past limits. The pages we cite above are the current source of truth.

If a limit matters to your work, check the usage page inside the product. Claude shows it under Settings, then Usage, and OpenAI points to the usage dashboard for current limits and reset times.