AI usage limits compared: 34 plans, and only Perplexity Free prints a number for its best mode

What are the usage limits of every major AI plan?

We read the pricing and help pages of nine vendors on 18 October 2026: OpenAI, Anthropic, Google, xAI, Perplexity, Microsoft, DeepSeek, Meta and Mistral. Then we sorted each plan by what its vendor says about the best model or mode it can use in chat. Side tools are left out. OpenAI does print a Codex estimate of 15 to 160 GPT-6.1 Sol messages per five hours on Plus, but it says "These estimates are not fixed message limits", and Codex is a coding tool. About the chat itself, most vendors say very little.

Every cell below is the vendor's own wording, from its pricing page or help center, in US prices, as of 18 October 2026. "Not published" means we found no figure or rule on any page the vendor runs. Flagship means the best chat model the plan can use.

Three of these vendors changed their plans between August and October 2026. xAI replaced SuperGrok with The X Subscription on 12 October 2026, and its announcement says "Nothing changes for existing SuperGrok, Cursor, or X Premium subscribers." Google put Gemini 4 Argon into AI Pro and Ultra on 13 October. Mistral renamed Le Chat to Vibe, and its help note of 12 August says your plan moves with you.

Which AI has the highest usage limit?

Claude Max 20x has the largest stated multiple over a free plan: at least 100 times Claude Free, for $200 a month. That's our own arithmetic, and it rests on two numbers that Anthropic prints. Its pricing FAQ says "Pro gives you at least 5x more usage per 5-hour session than Free", and Max 20x is 20 times Pro. Five times twenty is a hundred.

Google's chain is shorter. AI Ultra at $199.99 is 20 times AI Pro. AI Pro is 4 times a Gemini account with no plan, so Ultra comes to 80 times free. xAI's Grok Ultra at $200 is 50 times the $8 Plus plan. But xAI never ties Plus to the free plan, so its chain stops there. OpenAI prints no multiple at all. All it says is that Pro 500 "offers the highest included usage of the three plans."

These multiples don't cross from one vendor to the next. A unit of Claude usage isn't a unit of Gemini usage, and neither company tells you what its base is. Anthropic's usage help page says the budget moves with conversation length, the features you use, the model and "the effort level you've selected". So a ranking is only honest when you keep it inside one vendor.

Plain text chat is the exception, and ChatGPT wins it outright. OpenAI's pricing page lists everyday text chats as "Unlimited*" on Free, Go, Plus and Pro. The asterisk reads "Usage must be reasonable and comply with our policies." On Free and Go that chat runs on GPT-6 Luna. GPT-6.1 Sol, the flagship, starts on Plus.

Does ChatGPT or Claude have higher usage limits?

ChatGPT has the higher limit on everyday chat, and Claude has the clearer clock. OpenAI calls text chat unlimited on every plan it sells. Anthropic meters each Claude plan, and that includes Free, on a five hour window. Its pricing FAQ says "Every plan has usage limits that reset on a rolling five-hour session window, and paid plans add weekly limits on top."

The gap closes on the flagship. ChatGPT Plus gets GPT-6.1 Sol with "Expanded messages and uploads", and OpenAI's Plus help article adds that Plus "may include usage limits such as message caps, especially during high demand." Claude Pro gets Claude Opus 5.5 at "at least 5x" the Free allowance, and Anthropic says plainly that "there's no fixed message count". Neither of them gives you a number.

The two of them part ways at the wall. OpenAI says "ChatGPT may continue with another available model", so a Plus chat keeps going on something smaller. Claude stops, unless you've turned on usage credits, which run "at standard API rates". The full rules for each are in ChatGPT Plus limits and Claude Pro limits. If you're looking at a wall right now, Claude usage limit reached covers the reset times.

Pick by what hurts you most. If you send a lot of short questions, ChatGPT Plus won't stop you. If you need Claude Opus or Claude Fable for long work and want to know when you're back, Claude's five hour window tells you.

How often do AI usage limits reset?

AI usage limits reset every five hours on Claude and Gemini, weekly on Grok, Perplexity Pro and the ChatGPT Pro models, and daily on free Copilot, Meta AI and Perplexity Free. Anthropic and Google are the only two vendors on the five hour clock. Google's Gemini limits page says "Your limit refreshes every 5 hours until you reach your weekly limit." That's the same design as Anthropic's.

The weekly clocks don't share a day. Anthropic's Max plan article says the weekly limit "resets at a fixed time each week that is assigned to your account." xAI says its pool "resets every week on a schedule shown in the Usage tab in Settings." Perplexity gives the period and nothing more. So two people on the same plan can reset on two different days, and neither of them is wrong.

Free Copilot puts its clock in the error message itself. A free user on Microsoft's Q&A forum quoted the message on 4 August 2026: "You've reached your daily chat limit", then "check back tomorrow after" a time. The poster saw 7:42 PM once and a wait of more than a day another time. How Grok's pool and Gemini's weekly cap behave is in SuperGrok limits and Gemini usage limit reached.

What happens when you hit an AI usage limit?

One of four things happens at an AI usage limit: a smaller model takes over, you pay to keep going, you switch modes, or you wait. Which of them you get depends on the vendor, and sometimes on the plan you are on.

Google and OpenAI both hand you a smaller model. Google says "If you have a Google AI subscription and reach your limit, you can continue your conversation with Flash-Lite." OpenAI's model help article says that on Pro 200, "when you reach your GPT-6 Pro weekly limit, ChatGPT automatically switches to GPT-5.6" Thinking. On Pro 100, Pro 500 and managed work accounts, the model you were using "may be temporarily unavailable until the allowance resets." Free and Go text chat has no stated wall.

Anthropic and xAI will sell you more. On a paid Claude plan you can turn on usage credits, and they bill at API rates. xAI's Grok FAQ says "When your weekly usage limit is met, paid features will pause until your weekly limit resets." You keep the free Chat and Voice limits, and you can buy Extra Usage Credits on the web from $5. xAI warns that those credits are "priced at standard rates, which means the cost per action is higher" than your plan's.

Perplexity lets you switch to another mode. Its Pro help article says "Best" mode "is available without quota limits", so a Pro Search wall never stops searching. Perplexity Pro limits breaks down the other pools.

Everyone on a free plan has to wait. That is true of Claude Free, Gemini with no plan, Grok Free, free Copilot and Meta AI, whose help page says to "wait and try again later when your free limit refreshes." Don't ask support to lift it for you. OpenAI says "OpenAI Support does not reset ChatGPT or Codex usage limits." Microsoft's plans follow one rule: "If you reach a feature limit or use all of your AI credits, that feature may be unavailable until your allowance resets." The four ChatGPT limit messages decodes the wording on each one.

Why do trackers show message counts the vendors don't?

Tracker sites estimate their message counts or collect them from users, and none of the counts below is on a vendor page as of October 2026. AI Plan Finder, checked 29 July 2026, lists ChatGPT Plus at 160 messages per 3 hours and Claude Pro at about 45 messages per 5 hours. It then multiplies 45 by Anthropic's 5x and 20x to get about 225 and 900 for Max, and it says that it did. The AI Career Lab, updated 11 October 2026, labels its Claude counts as user reported.

OpenAI's own article on the current models, GPT-6 and other models in ChatGPT, prints no message count for any plan. The old URL for its GPT-5 article now opens that same page, too. So a "160 every 3 hours" you read today describes an older model or one person's account. Treat any count as an estimate. Trust the reset time your own app shows.

Which plan should you pick when the limit is the problem?

Pick the plan whose wall you can live with, because only one of the 34 gives you a number for its best mode, and that one is free. If you want volume for everyday chat, ChatGPT Free, or Go at $8, already says it is unlimited. If you want a clock you can see, Claude and Gemini refill every five hours. If it is search you need, Perplexity Pro's Best mode has no quota. If you want a printed count, Perplexity Free's 3 Pro Searches a day is the only one, and it's small.

Whizi, which publishes this page, prints a price per message instead of a multiple. Whizi Pro is $29.99 a month for 2,500 credits, and each model's credit cost is fixed: Claude Opus is 20 credits a message and Claude Sonnet is 10, so the plan buys 125 Opus or 250 Sonnet messages a month. Every Whizi plan is capped at 10 messages a minute and 60 an hour, and credits reset when the UTC calendar month turns over. On raw volume Whizi loses to ChatGPT Free, which charges nothing for unlimited everyday chat. It fits you if you want GPT, Claude and Gemini on one bill, with a count you can plan around.