Anthropic reshuffled its Claude plans at the end of October 2026. The old $200 “Max 20x” plan is now called Claude Ultra, verified Ultra accounts lose the five-hour session limit entirely, and both Max and Ultra now come with monthly API credits. If you have been staring at the pricing page wondering whether Pro is enough or whether Max is worth five times the money, this guide is for you.
I checked every number below against Anthropic’s own pricing page (claude.com/pricing) and its help centre article on usage limits on November 5, 2026.
- Free ($0): Sonnet and Haiku, limited usage every 5 hours and every week, up to 5 projects, no Claude Code.
- Pro ($20/month, or $17/month billed yearly at $200): Opus, Claude Code, Research, Claude in Chrome, Fable only through paid usage credits.
- Max ($100/month): 5x Pro usage, Fable included up to half your weekly limit, $100/month in API credits, priority access.
- Ultra ($200/month): 10x Pro weekly usage, no 5-hour limit for verified accounts, $200/month in API credits.
- Chat and Claude Code share the same limits. When you run out, optional usage credits let you keep going at API rates.
Claude Plans and Prices at a Glance
| Plan | Price (US, before tax) | Limit every 5 hours | Limit every week | Monthly API credits |
|---|---|---|---|---|
| Free | $0 | Limited | Limited | No |
| Pro | $20/month, or $200/year ($17/month) | 1x (baseline) | 1x | No |
| Max | $100/month | 5x Pro | 5x Pro | $100/month |
| Ultra | $200/month | No 5-hour limit (verified accounts) | 10x Pro | $200/month |
Prices on the iOS and Android apps can be different because of app store fees, and the price shown on claude.com excludes tax. For teams there are separate Team and Enterprise plans with central billing, SSO and admin controls, which none of the individual plans include.
Feature Comparison Table
Most of the core features are now on every plan, including the free one. The differences are mostly about which models you can pick, how much you can use them, and the work tools aimed at professionals.
| Feature | Free | Pro | Max | Ultra |
|---|---|---|---|---|
| Chat on web, desktop, iOS, Android | Yes | Yes | Yes | Yes |
| Web search, voice mode, memory | Yes | Yes | Yes | Yes |
| Artifacts, Claude Design, Slides, Docs | Yes | Yes | Yes | Yes |
| Connectors, Skills, desktop extensions | Yes | Yes | Yes | Yes |
| Create and edit files with code execution | Yes | Yes | Yes | Yes |
| Projects | Up to 5 | Unlimited | Unlimited | Unlimited |
| Sonnet and Haiku models | Yes | Yes | Yes | Yes |
| Opus models | No | Yes | Yes | Yes |
| Fable | No | Usage credits only | Up to 50% of weekly limit | Up to 50% of weekly limit |
| Claude Code | No | Yes | Yes | Yes |
| Research mode | No | Yes | Yes | Yes |
| Claude in Chrome, Microsoft 365, Outlook | No | Yes | Yes | Yes |
| Hand off and schedule tasks | No | Yes | Yes | Yes |
| Claude Dashboards (beta), Claude Science | No | Yes | Yes | Yes |
| Priority access at busy times | No | No | Yes | Yes |
| Context window | Up to 1M tokens on every plan, depending on the model | |||
If you mainly want a smart assistant for questions, writing and the occasional file, the free plan is surprisingly complete. The jump to Pro is about Opus, Claude Code and Research. The jump to Max or Ultra is almost entirely about usage volume and Fable.
How Claude Usage Limits Work
Claude does not count messages. It counts how much compute your conversations use, so the number of messages you get depends on the model, how long the chat is, how big your files are and which features you turn on. Anthropic’s help centre explains the system like this:
- A session limit that resets every five hours. Your five-hour window starts with your first message. Free, Pro and Max have it. Verified Ultra accounts do not.
- A weekly limit on every paid plan that resets once a week. Max and Ultra also have a separate weekly limit for Fable.
- Claude chat and Claude Code share the same pool. In Anthropic’s words, your combined usage across both interfaces counts toward your limits. A long coding session in the terminal leaves less for chat, and the other way round.
- Some features burn faster. Research mode runs many searches and analysis steps, and project files count toward usage whenever they are part of the conversation.
You can see where you stand in Settings > Usage, which shows a live bar for the current session and the week, plus any usage credit spend for the month.
Fable Access by Plan
Claude Fable 5.1 is Anthropic’s most capable widely released model, and it is expensive to run (on the API it costs $10 / $50 per million tokens, against $4 / $20 for Opus 5.5). That is why the plans treat it differently:
- Free: no Fable.
- Pro: Fable is available, but only paid for with usage credits, so every Fable message comes out of your pay-as-you-go balance rather than your plan allowance.
- Max and Ultra: Fable is included, up to half of your weekly limit. Once you hit the Fable cap you can keep using Opus, Sonnet and Haiku for the rest of the week.
For most everyday work, Opus 5.5 is the better default anyway. It is faster, it is the model Anthropic recommends for most workloads, and it currently tops the Artificial Analysis Intelligence Index. Save Fable for long, hard reasoning or research jobs.
What Changed With Claude Ultra
Ultra is the old $200 Max 20x plan with a new name and a few changes. According to MIXED News (November 1, 2026), which reported Anthropic’s October 30 announcement:
- Existing Max 20x subscribers moved to Ultra automatically at the same $200 price.
- Verified Ultra accounts have no five-hour session limit. Anthropic verifies accounts automatically using signals such as account and payment history. Until verification finishes, a temporary limit applies, and adding a phone number at checkout does not remove it on its own.
- A verified account no longer shows the “Current session” bar in Settings > Usage. That is the quickest way to check your status.
- The weekly limit stays, at 10x Pro, plus a separate Fable weekly limit. Anthropic can still add limits for capacity and fair access.
In practice this matters most for people who run Claude Code for long stretches. On Max and Pro, a heavy morning can lock you out until the five-hour window resets. On a verified Ultra account you only have to watch the weekly total.
Usage Credits vs Monthly API Credits
These two sound alike but are completely separate, and mixing them up is a common source of confusion.
| Usage credits | Monthly API credits | |
|---|---|---|
| Plans | Pro, Max, Ultra | Max ($100), Ultra ($200), Team |
| What they pay for | Extra Claude app and Claude Code usage after you hit a limit, plus Fable on Pro | Your own apps built on the Claude Platform (API) |
| Price | Billed at standard API rates | Included in the subscription |
| Where to manage | Settings > Usage (web only for mobile subscribers) | Claude Console |
Anthropic’s help article is explicit that the monthly API credits on Max and Ultra cannot be used to pay for usage credits. If you build apps, see my Claude API pricing guide to work out how far $100 or $200 a month goes. As a rough guide, $100 buys about 25 million Opus 5.5 input tokens, or about 1 billion Haiku 5.5 input tokens on short prompts.
When usage credits are on and you reach a limit, your requests simply continue on pay-as-you-go instead of stopping. Spending continues until you hit your own monthly spend limit, or your balance runs out if auto-reload is off. Usage credits do not change when your limits reset.
How to Make Your Limit Last Longer
- Pick the smallest model that does the job. Sonnet 5.5 and Haiku 5.5 use far less of your allowance than Opus or Fable.
- Start new chats often. Every message re-reads the whole conversation, so a 200-message thread costs much more per reply than a fresh one. Ask Claude for a summary and paste it into a new chat.
- Use Projects for reference files rather than re-uploading the same PDFs, and remove files you no longer need.
- Save Research mode for real research. It runs many searches per question.
- In Claude Code, keep context lean: clear or compact the conversation between tasks and lower the effort level for routine edits.
- Watch Settings > Usage before a big task so you do not get cut off halfway.
If you are getting error messages rather than limit warnings, the cause may be an outage or overload instead. My guide to Claude errors 429, 529 and “overloaded” explains each one.
Which Claude Plan Should You Buy?
| You are… | Best plan | Why |
|---|---|---|
| Trying Claude, light questions and writing | Free | Sonnet, web search, files, Artifacts and memory at no cost |
| Daily writer, student or office user | Pro | Opus, Research, Projects, Claude in Chrome and Microsoft 365 |
| Developer using Claude Code a few hours a day | Pro, then Max if you hit limits | Start cheap; upgrade when the 5-hour window keeps stopping you |
| Developer who also builds on the API | Max | $100 of API credit roughly pays for the upgrade from Pro |
| All-day agent or Claude Code user | Ultra | No 5-hour limit once verified, 10x weekly usage, $200 API credit |
Comparing Claude with ChatGPT, Gemini or Grok subscriptions? I put them side by side in ChatGPT vs Claude vs Gemini vs Grok, and the ChatGPT tiers have their own breakdown in ChatGPT Free vs Go vs Plus vs Pro.
Claude Team and Enterprise Prices
If two or more people need Claude for work, the business plans are often better value than separate Pro accounts, because they add admin controls, central billing and no training on your data by default. These are the prices on Anthropic’s pricing page (November 5, 2026, before tax):
| Plan | Billed yearly | Billed monthly | Seats | Notes |
|---|---|---|---|---|
| Team Standard seat | $20/seat/month | $25/seat/month | 2 to 150 | Pro-level usage per person |
| Team Premium seat | $100/seat/month | $125/seat/month | 2 to 150 | 5x the usage of a Standard seat |
| Enterprise | $20/seat/month plus usage at API rates | Annual only | 20 minimum | SSO, invoicing via sales |
Team plans also include up to $500 a month in pooled API credits, and you can mix Standard and Premium seats, so only your heaviest users need the expensive seat. Education plans for whole institutions exist but are priced through sales.
Yearly Cost of Each Claude Plan
| Plan | Per year | Per day | API credits included per year |
|---|---|---|---|
| Pro (monthly) | $240 | About $0.66 | None |
| Pro (yearly) | $200 | About $0.55 | None |
| Max | $1,200 | About $3.29 | $1,200 |
| Ultra | $2,400 | About $6.58 | $2,400 |
The API credits change the maths for developers. If you would spend $100 a month on the Claude API anyway, Max effectively gives you 5x Pro usage in the app for free. If you never touch the API, Max costs five times Pro for five times the usage, which is only worth it if you hit Pro’s limits most weeks. Yearly Pro saves $40 compared with paying monthly.
Claude Pro vs ChatGPT Plus vs Google AI Pro
| Claude Pro | ChatGPT Plus | Google AI Pro | |
|---|---|---|---|
| Price | $20 ($17 yearly) | $20 (monthly only) | $19.99 |
| Top model | Opus 5.5 | GPT-6 Astra | Gemini 4 Argon |
| Context | Up to 1M (model dependent) | 256K reasoning | Up to 1M on Pro models |
| Coding agent | Claude Code included | Codex included | Gemini CLI and Jules |
| Image generation | No native generator | Yes | Yes (Nano Banana) |
| How limits work | 5-hour session plus weekly | Rolling per-model caps | Daily caps per feature |
The short version: pick Claude for coding, writing and long documents, ChatGPT for an all-in-one assistant with images and voice, and Gemini if you live in Gmail and Docs. More detail in ChatGPT vs Claude vs Gemini vs Grok and ChatGPT plans compared.
How to Upgrade, Cancel or Get a Refund
Upgrade
- Sign in at claude.ai and click your name or initials (bottom left).
- Choose Upgrade plan (or Settings > Billing if you already pay).
- Pick Pro, Max or Ultra, and monthly or yearly for Pro. Prices on claude.com exclude tax, which is added at checkout.
Subscribing on the web usually costs less than in the iOS or Android app, where app store fees can raise the price.
Cancel
- Go to Settings > Billing on claude.ai.
- Click Cancel (or Adjust plan to move down a tier).
- Cancel at least 24 hours before renewal to avoid the next charge.
Your plan stays active until the end of the billing period, and cancelling does not delete your chats or projects. If you subscribed through the App Store or Google Play, cancel in that store instead. Team owners cancel in Organization settings.
Refunds
Anthropic says payments are generally non-refundable, with exceptions under its Consumer Terms or local law. In the EEA and UK you can request a refund in the app within your 14-day withdrawal period. Elsewhere, use the in-app refund request to check eligibility. App store purchases are refunded by Apple or Google.
Privacy: Does Claude Train on Your Chats?
On Free, Pro, Max and Ultra you choose whether your chats can be used to improve Claude. Check the setting in Settings > Privacy (“Help improve Claude”) and turn it off if you do not want that. Incognito chats are not saved to your history. Team and Enterprise plans are covered by Anthropic’s commercial terms, which do not use your data for training by default. If you work with client data, use a Team plan rather than a personal one.
Limit and Billing Problems (Fixes)
“You’ve reached your usage limit” on Pro
Check Settings > Usage to see whether it is the 5-hour session limit or the weekly limit. A session limit clears when the window resets. For a weekly limit you can wait, turn on usage credits to keep going at API rates, or upgrade. Starting a new chat instead of continuing a very long one also helps.
Ultra still shows a 5-hour limit
Your account has not finished verification. Anthropic verifies automatically from account and payment history, and adding a phone number alone does not speed it up. When verification completes, the “Current session” bar disappears from Settings > Usage.
Claude Code says I am out of usage but chat still works (or the other way round)
Both share one pool, but different models drain it at different rates. Switch Claude Code to Sonnet for routine edits and keep Opus for hard problems.
My card was declined
Make sure the billing address and country match the card, turn off VPNs, and check that your bank allows international online payments. Try a different card if the problem continues.
For error codes from the API or Claude Code, such as 429 or 529 overloaded, see my Claude errors fix guide.
Frequently Asked Questions
How many messages do I get on Claude Pro?
Anthropic does not publish a fixed message count. Limits are based on usage, so short chats with Sonnet go much further than long chats with Opus and large files. You get a session allowance every five hours and a weekly cap.
When does the Claude limit reset?
The session limit resets five hours after it starts, and the weekly limit resets once a week. Settings > Usage shows the exact reset times for your account.
Does Claude Code use my chat limit?
Yes. Claude conversations and Claude Code draw from the same limits on every paid plan.
Is Claude Max worth it?
If you keep hitting Pro limits or you also pay for the API, yes. Max gives 5x the usage, included Fable, priority access and $100 a month in API credits for $80 more than Pro.
Is there a student discount?
The pricing page does not list one for individual plans. The annual Pro plan ($200 a year) is the cheapest way to pay for Pro.
Can I use Claude for free?
Yes. The free plan includes Sonnet and Haiku, web search, voice, memory, connectors and up to five projects, with limits every five hours and every week.
How much is Claude Pro per year?
$200 if you pay yearly (about $17 a month), or $240 if you pay $20 monthly. Tax is added on top.
Can I get a refund on Claude Pro?
Generally no, except where the law requires it. EEA and UK customers can ask for a refund within 14 days. Others can check eligibility through the in-app refund request.
What is the difference between Claude Max and Claude Ultra?
Max ($100) gives 5x Pro usage with a 5-hour session limit. Ultra ($200) gives 10x Pro weekly usage, removes the 5-hour limit for verified accounts, and includes $200 a month of API credits instead of $100.
Is Claude Team cheaper than Pro?
A Team Standard seat costs the same $20 a month on yearly billing (or $25 monthly) with a two-seat minimum, and adds admin controls and pooled API credits.
Does cancelling Claude delete my chats?
No. Cancelling keeps your chats and projects; you drop to the Free plan at the end of the billing period.
Final Thoughts
For most people, the free plan or Pro is plenty. Pro is the sweet spot for anyone who wants Opus, Claude Code and Research without thinking too hard about limits. Max makes sense once limits start to bite or once you can use the $100 of API credit, and Ultra is now a genuinely different product for people who run Claude all day, because removing the five-hour window changes how you work.
If you would rather run models on your own hardware with no limits at all, have a look at my best Ollama models by VRAM list and the DeepSeek vs Claude comparison.




