What are the Claude Pro limits, plan by plan?
Claude Pro's limits are a five-hour session allowance of at least 5x Free and a weekly cap with no published size. Around those two sit hard numbers: up to 1M tokens of context, 500 MB per uploaded file, and Fable only through paid usage credits. We compiled every row below from claude.com/pricing and eight Claude Help Center articles on 18 October 2026. It covers the four individual plans, priced in US dollars.
| Limit, as of 18 October 2026 | Free | Pro | Max 5x | Max 20x | What Anthropic prints |
|---|---|---|---|---|---|
| Price a month | $0 | $20, or $17 billed annually | $100 | $200 | A number |
| Usage per five-hour session | "Everyday questions" | At least 5x Free | 5x Pro | 20x Pro | A multiplier, no size |
| Session window | Rolling five hours | Rolling five hours | Rolling five hours | Rolling five hours | A number |
| Weekly limit | None stated | Yes, reset at a fixed time set per account | Yes | Yes | No size at all |
| Claude Opus | No | Yes | Yes | Yes | Yes or no |
| Claude Fable | No | Usage credits only | Up to 50% of the weekly limit | Up to 50% of the weekly limit | A number |
| Claude Code | No | Yes, same limits as chat | Yes | Yes | Yes or no |
| Research | No | Yes, same limits as chat | Yes | Yes | Yes or no |
| Context window in chat | Up to 1M tokens, varies by model | 1M on the newest models, 500K or 200K on older ones | Same as Pro | Same as Pro | A number |
| File uploads in chat | 500 MB a file, 20 files a chat | Same | Same | Same | A number |
| Projects | Up to 5 | No cap printed; 30 MB a project file | Same as Pro | Same as Pro | A number |
| Past the limit | Wait or upgrade | Usage credits at API rates; bundles up to $2,000 a month | Same as Pro | Same as Pro | A number |
| Monthly API credits | No | No | $100 | $200 | A number |
| Output limit | Not stated | Not stated | "Higher" | "Higher" | A word |
| Priority at busy times | No | No | Yes | Yes | Yes or no |
Count the last column. Eight of the 15 rows carry a number, four are a yes or no, and three carry no size at all. Those three are the ones you'd buy a plan for: how much you get per session, how much per week, and how long Max's replies can run. Anthropic's Pro plan page says why there's no count: "The number of messages you can send will vary based on message length, including the length of files you attach". The same sentence adds the length of the conversation and the model or feature you pick. The upload row has no plan split because the upload help page gives none.
How much usage does Claude Pro give per five hours?
Claude Pro gives at least five times the Free allowance per five-hour session, and Anthropic doesn't say what the Free allowance is. The claude.com FAQ puts it this way: "Pro gives you at least 5x more usage per 5-hour session than Free." The Max plan page stacks two more multipliers on top, five and twenty times Pro's per-session allowance. Multiply them out and the ladder is long.
Divide by price and one plan stands out. Pro buys one Free session's worth of allowance per $4. So does Max 5x, at $100 for 25. Max 20x buys one per $2, because it's four times Max 5x for twice the price. Treat these as floors: Pro is "at least" 5x, and every Max multiple is defined against Pro. None of it is a count.
The last time Anthropic put an absolute figure on any of this was July 2025. Its email to subscribers, reported by TechCrunch on 28 July 2025, said most Pro users could expect 40 to 80 hours of Sonnet 4 a week in Claude Code, and the $200 Max plan 240 to 480. Those hours measured Claude Code on Sonnet 4, a model that no longer appears in Anthropic's context window table. Anthropic has changed limits at least six times since, and it hasn't restated the hours.
Does Claude Pro have a weekly limit?
Yes. Claude Pro and both Max plans have a weekly usage limit on top of the five-hour session, and the Free plan has none that Anthropic states. The pricing FAQ says every plan resets "on a rolling five-hour session window, and paid plans add weekly limits on top." On Pro the weekly limit "applies across all models" and resets "at a fixed time each week that is assigned to your account." That day and time don't move when you subscribe or when you start chatting. Free has no stated weekly cap.
One pool covers everything. Anthropic's usage and length limits page says your use of every Claude surface counts towards the same limit: claude.ai, the desktop and mobile apps, and Claude Code. Max adds a second weekly bar for Fable. Settings > Usage shows each limit as a progress bar, never as a count, with the time left in the current session beside it. For when each bar refills, and what a saved limit reset does to it, see when Claude limits reset.
The Pro page keeps a door open for more caps. Anthropic writes that "we may limit your usage in other ways, such as weekly and monthly caps or model and feature usage, at our discretion." No monthly cap is in force for Pro as of October 2026. The sentence lets Anthropic add one without changing the plan.
Can you use Opus and Claude Code on Claude Pro?
Yes to both: Claude Pro includes Claude Opus and Claude Code, and the Free plan includes neither. Opus and Claude Code draw on the same session and weekly limits as chat. Anthropic's Claude Code plan page says "all activity in both tools counts against the same limits", IDE use included. An afternoon in the terminal is gone from the chat window too.
Fable is the exception on Pro. Anthropic's Fable plan page says Fable 5 and 5.1 "run on pay-as-you-go usage credits" on Pro, outside the plan's limits. Max includes them, and up to half of a Max weekly limit can go to Fable before credits take over. Research works on every paid plan and "is subject to the same limits as standard Claude conversations", per its help page. It also warns that a research run can use them up faster.
Context is the same on Pro and Max. Anthropic's context window page gives 1M tokens in chat to Fable 5.1, Opus 5.5, Opus 5, Sonnet 5.5, Sonnet 5 and Haiku 5.5 on every paid plan, as of October 2026. Opus 4.6 to 4.8, Sonnet 4.6 and Fable 5 get 500K, and older models 200K. Projects stop at 5 on Free. Paid plans get no printed project cap, with each project file capped at 30 MB.
What happens when you hit the Claude Pro limit?
On Free you wait for the five-hour window or upgrade. On Pro and Max you can also switch on usage credits, which Anthropic bills "at standard API rates" once a session, weekly or Fable limit is spent. Nothing charges until then. You prepay, set a monthly spend cap or leave it unlimited, and the wall notice changes to "Now using credits." Anthropic's bundles page sells credits at a discount: $50 of usage for $45, $250 for $200 and $1,000 for $700, up to $2,000 a month.
Two catches sit in the fine print. Max's $100 or $200 of monthly API credits "can't be used in Claude Code or the Claude apps, including for extra usage", so they don't extend the plan you're on. And in Claude Code the prompt cache lasts an hour on a subscription but drops to five minutes once you're drawing on usage credits, per Anthropic's cost docs. A break of a few minutes then costs a full re-read of the context.
Anthropic also hands out free limit resets, "given occasionally to eligible plans", which refill either the session or the week when you choose to use one. If you're looking at the wall right now, Claude usage limit reached walks through the message and the fastest fix.
How did Claude's limits change in 2026?
Anthropic changed Claude's limits six times between March and September 2026, and four of the six raised a limit it has never sized. The other two cut: a peak-hours squeeze on five-hour sessions in March, and Fable 5 moving off Pro's allowance in July. Every row but March is from Anthropic's own news posts or help center; March came from an Anthropic staff post that PCWorld reported on 27 March.
| Date | What changed | Plans | Source |
|---|---|---|---|
| Late March 2026 | Five-hour session limits tightened on weekdays from 5 to 11 a.m. PT, weekly limits unchanged; Anthropic estimated about 7% of users would hit limits they hadn't before | Free, Pro, Max | Anthropic staff post, via PCWorld |
| 6 May 2026 | Claude Code five-hour limits doubled; the peak hours reduction on Claude Code removed | Pro, Max, Team, seat-based Enterprise (peak hours: Pro and Max) | Anthropic news |
| 13 May to 13 September 2026 | Claude Code weekly limits 50% higher, five-hour limits unchanged | Pro, Max, Team, legacy seat-based Enterprise | Claude Help Center |
| 19 July 2026 | A promotion letting Fable 5 use up to 50% of the weekly limit at no extra cost ended; on Pro, Fable 5 now runs on usage credits only | Pro, Team standard seats | Claude Help Center |
| 14 September 2026 | Claude Code weekly limits settle 25% above the pre-May level | Same plans | Claude Help Center |
| 22 September 2026 | Five-hour limits raised with Opus 5.5, size not given; a limit reset to save and use later | Pro, Max, Team, seat-based Enterprise | Anthropic news |
The 6 May post opened with "we're doubling Claude Code's five-hour rate limits for Pro, Max, Team, and seat-based Enterprise plans." The Opus 5.5 post of 22 September said Anthropic is "increasing five-hour usage limits" with no figure. The promotion page still carries "may-july" in its address, though it ran to 13 September. The Fable plan page dates the Fable change to 11:59 PM PT on 19 July. Free users got none of the raises.
What if Claude Pro's limits are too low for you?
Claude Pro's limits go further if you follow Anthropic's own advice before paying for a bigger plan. Its usage and length limits page says to start a new conversation when a long chat nears the limit, turn extended thinking off when you don't need it, and pick a lower effort level for routine work. It also says to switch off connectors and web search a chat doesn't use. Its best practices page adds that documents kept in a Project are cached and count less when you reuse them. That cache expires after a long break, though.
If the bar still runs out, Anthropic's advice on the Claude Code page is to move from Pro to Max 5x "if you consistently hit limits", and from Max 5x to Max 20x on the same signal. Watch the weekly bar in Settings > Usage for a week before you pay more. If it passes half by midweek, the bigger plan is buying you time. Claude Pro vs Claude Max works through who needs the multiplier.
If the problem is that no plan prints a number, a meter that does is the other route. Whizi, which publishes this page, sells Pro at $29.99 a month for 2,500 credits. Claude Sonnet costs 10 credits a message and Claude Opus 20. That's 250 Sonnet or 125 Opus messages a month, whatever each message holds (how credits work). It loses on things Anthropic's plans do well. Claude Code still needs a Claude plan or an API key, a turn carries at most 40,000 input tokens, every plan stops at 60 messages an hour, and credits refill only when the UTC month turns. It suits someone who wants to know the count before sending, not someone living in the terminal.
- Open Settings > Usage and note which bar is fuller: the five-hour session or the week
- On Pro, expect Fable to charge usage credits from the first message; on Max, up to half the weekly limit covers it
- Count Claude Code and Research against the same pool as chat; an afternoon in the terminal is missing from Thursday's chats
- When you enable usage credits, set a monthly spend limit with "Adjust limit" instead of picking "Set to unlimited"
- Don't expect Max's monthly API credits to extend the plan; they can't be used in Claude apps or Claude Code
Frequently asked questions
How many messages do you get with Claude Pro per day?
Anthropic doesn't publish a daily or per-session message count for Claude Pro. It says Pro gives at least 5x Free's usage per five-hour session, that the count varies with message length, attachments, conversation length and model, and that a weekly cap sits on top. The session refills five hours after it starts, so a long day can span several windows until the weekly bar fills. Is Claude Pro worth it runs the numbers on what that buys.
What are Claude Pro limits vs the free plan?
Claude Pro gives at least five times the Free plan's usage per five-hour session, and adds a weekly cap Free doesn't have. Pro also adds Claude Opus, Claude Code, Research, Projects with no printed cap instead of 5, and usage credits past the limit, for $20 a month or $17 billed annually. Both plans share the same context window and upload limits. Is Claude Free good enough covers who can stay on Free.
Does Claude Max have Opus limits?
Yes, but Anthropic doesn't size it: Opus draws on Max's five-hour and weekly limits, and Claude Code's error docs also list a separate "You've hit your Opus limit" that applies only to Opus requests. Switching to a model outside the Opus family with /model keeps you working after that message. The one model with a printed cap on Max is Fable, which can use up to 50% of the weekly limit before usage credits take over. Is Claude Max worth it weighs the $100 and $200 tiers.
Are Claude Pro limits counted in tokens?
Yes, in effect: Claude Pro meters usage by how much text Claude processes, so a long thread or a large file spends the allowance faster than short messages do. Every turn re-reads the whole conversation, which is why Anthropic says longer chats consume more of the limit. Project files are cached and count less when reused. The separate wall for a chat that's too long is the context window, covered in Claude conversation too long.
