Claude Pro vs Max, and Sonnet vs Opus
Which Claude plan to pay for, and which model to pick for each kind of work. Short version: most people need Pro and Opus; Max and Ultra are for daily heavy use.
We have no affiliate deal with Anthropic, and nobody paid to appear on this page.
- use Claude every day for writing, research and analysis
- want Opus, Research, Claude Code and Claude in Chrome
- hit the limit a few times a month, not most days
- want yearly billing: $17 a month, $200 up front
- run Claude Code or long agent tasks most days
- hit the five-hour limit regularly on Pro
- want Fable without paying per token
- need priority access when Claude is busy
Which model: Anthropic tells developers to start with Opus 5.5 for most work 5. In the app, use Sonnet 5.5 for quick drafts and lookups, Opus 5.5 for anything you will act on, and Fable 5.2 only for long, multi-step jobs where Opus falls short.
Claude plans at a glance
| Plan | Price | Models | Usage | Best for |
|---|---|---|---|---|
| Free | $0 | Sonnet, Haiku 1 | Baseline | Trying Claude; up to 5 projects |
| Pro | $20, or $17 billed yearly 1 | Opus, Sonnet, Haiku; Fable on credits | More than Free | Daily individual use |
| Max | $100, monthly only 3 | All, Fable up to half the weekly limit 7 | 5x Pro, per session and per week | Daily Claude Code or agent work |
| Ultra | $200, monthly only 14 | All, Fable up to half the weekly limit 1 | 10x Pro per week, no five-hour limit | All-day heavy use |
| Team Standard | $25, or $20 billed yearly 4 | Opus, Sonnet, Haiku; Fable on credits | 1.25x Pro | Teams of 2 to 150 with shared projects |
| Team Premium | $125, or $100 billed yearly 4 | All, Fable up to half the weekly limit | 6.25x Pro | Heavy users inside a team |
| Enterprise | $20 a seat a month, billed yearly, plus usage at API rates; 20 seats minimum 1 | All; on usage-based plans Fable is billed at API rates 7 | Billed by consumption | Companies that need SSO, SCIM and audit logs |
Pro vs Max and Ultra
The only reason to pay for Max or Ultra is more usage. Max includes everything in Pro with five times the usage, per session and per week 3. Ultra, which replaced the old Max 20x plan at the same $200, gives ten times Pro's weekly usage and drops the five-hour limit for verified accounts; Max 20x subscribers moved to it automatically 14. Anthropic adds higher output limits, early access to new features and priority access at peak times 1.
Pro already covers the main tools: Opus, Research, Claude Code, Claude in Chrome and Claude for Microsoft 365 1. If you rarely see a limit warning, Max and Ultra buy you nothing you will notice.
What Max and Ultra add
- More usage: five times Pro on Max; ten times Pro's weekly usage and no five-hour limit on Ultra 3 14.
- Fable included. Up to half of your weekly limit can go to Fable models at no extra cost; Ultra keeps a separate weekly limit for Fable 14. On Pro, Fable runs only on usage credits 7.
- Monthly API credits: $100 on Max and $200 on Ultra 1. These are for the developer API only; they cannot be spent in the Claude apps or Claude Code 11.
- Longer background tasks on reports, spreadsheets and presentations 3.
The catches
Max and Ultra are billed monthly only, while Pro has a yearly option at $200 1. Upgrades are prorated, and moving from yearly Pro to Max can leave you an account credit 3. Anthropic runs no standing discounts on Pro, Max or Ultra.
There is also a middle path. Pro subscribers can turn on usage credits and pay API rates only when they run past the limit, with a monthly spend cap 13. If you hit the limit a few times a month, that usually costs less than an extra $80.
Sonnet vs Opus
Opus 5.5 is the stronger model and Anthropic's suggested default; Sonnet 5.5 is faster and half the API price 5. On a subscription you don't pay per token, but the model you pick changes how fast you use up your limit 9.
| Model | Built for | Speed | API price |
|---|---|---|---|
| Fable 5.2 | Demanding reasoning and long-running agent work | Moderate | $8 / $40 5 |
| Opus 5.5 | Long-running agentic coding and knowledge work | Moderate | $4 / $20 |
| Sonnet 5.5 | The best mix of speed and intelligence | Fast | $2 / $10 |
| Haiku 5.5 | High-volume, quick tasks such as sorting and extraction | Fastest | From $0.10 / $0.50 |
All four read text and images, hold a 1 million token context window and share a June 2026 knowledge cutoff 5. The paid apps give all four the full 1M window 10.
Use Opus for
Work someone acts on without recheckingAnthropic's model guide points Opus at multi-hour coding agents, large refactors, vision-heavy work and computer use 6. In the app, that translates to contract reviews, financial summaries, client deliverables and anything with numbers. Research also runs on Opus 5.5 whichever model you have selected 12.
Use Sonnet for
Drafts you will edit anywaySonnet is Anthropic's pick for code generation, data analysis, content creation and agentic tool use where speed matters 6. It is a good default for first drafts, brainstorms, email replies and quick questions.
Our rule of thumb: sort each repeated task by one question, who checks the output? If a person reviews it before it goes anywhere, the faster model is fine. If nobody does, use Opus. The cost of a quiet mistake is higher than the cost of the better model.
Effort matters as much as the model. Anthropic notes that tuning effort is often a better lever than switching models 6. In the app, the model menu next to the send button sets both: Low and Medium for routine work, High for the best balance of quality and speed, Extra high for long coding and agent tasks, and Max for the hardest problems 8.
Fable and Haiku
Fable 5.2: the top model, with conditions
Fable 5.2 launched on 9 November 2026 16. Anthropic built it for demanding reasoning and long-horizon agent work, and it costs 20% less per token on the API than Fable 5.1 15. Its sweet spot is agent sessions that run for hours, multistep deep research, and analysis carried through to a finished document, spreadsheet or deck. Anthropic's own advice is to move to Fable only when Opus at Extra high or Max effort still falls short.
Access depends on your plan 7:
- Free: not available.
- Pro and Team Standard: outside the plan's limits; every Fable message runs on usage credits.
- Max, Ultra and Team Premium: included, up to half of your weekly limit. After that, pay with credits or switch models.
Pick Fable in the model picker on web, desktop or mobile. Fable uses your weekly limit faster than other models 7, so plan the task with Opus first and hand Fable a clear brief.
Haiku 5.5: fast and cheap
Haiku 5.5 is the fastest and cheapest Claude model, built for real-time apps, high-volume processing and helper agents 6. It is on every plan, Free included 1. In the app, use it for quick lookups, reformatting and short summaries when you want to save your limit for Opus.
How limits differ by plan
Paid plans have a session limit that resets every five hours and a weekly limit on top; verified Ultra accounts have no five-hour limit 9. Anthropic does not publish message counts; it describes each plan as a multiple of another:
- Pro gives more usage than Free, every five hours and every week 1.
- Max gives five times Pro's 3; Ultra ten times Pro's weekly usage, with no five-hour limit for verified accounts 14.
- Team Standard seats get 1.25 times Pro; Premium seats 6.25 times Pro, per member and not shareable between seats 4.
The weekly limit resets at a fixed time assigned to your account, and you can see it under Settings > Usage 2. The same limits cover Claude chat and Claude Code 11. For the details, see Claude usage limits, explained.
Team and Enterprise
Team needs at least two members and tops out at 150 seats 4. It adds shared projects, central billing, SSO, admin controls and connectors for Google Drive, Gmail, Calendar, GitHub, Microsoft 365 and Slack. Anthropic does not train on Team content by default 1. Premium seats cost five times as much as Standard and carry five times the usage, so they suit the people who live in Claude Code all day.
Enterprise starts at $20 a seat per month billed yearly, plus usage at API rates, with a 20-seat minimum. It adds SCIM, audit logs, custom data retention and a HIPAA-ready option 1.
Questions people ask
Is Claude Max worth it?
Only if you hit Pro's limit often. Max and Ultra add usage, Fable access and priority, not new everyday tools. If you run out a few times a month, Pro with capped usage credits is cheaper.
Is Opus better than Sonnet?
Opus 5.5 is the more capable model and Anthropic's default recommendation. Sonnet 5.5 is faster and uses your limit more slowly, which makes it better for drafts and quick questions.
Can I use Fable on Claude Pro?
Yes, but only through usage credits billed at API rates. Fable is included in the plan limits on Max, Ultra and Team Premium, up to half of the weekly limit.
Does Claude Pro include Claude Code?
Yes. Pro, Max and Ultra include Claude Code in the terminal, and it shares the same usage limits as chat.
What happened to Claude Max 20x?
It is now called Ultra: $200 a month, ten times Pro's weekly usage and no five-hour limit for verified accounts. Existing Max 20x subscribers moved over at the same price.
Can I pay for Claude Max yearly?
No. Max and Ultra are monthly only. Pro has a yearly option at $200, which works out to about $17 a month.
Which model does Claude Free use?
Free includes Sonnet and Haiku. Opus, Fable, Research and Claude Code need a paid plan.
Related
The next model change, in your inbox
195,000 readers get Uncover AI's practical AI briefing. We tell you what changed and whether it changes your pick.
Sources
- Claude pricing
- What is the Pro plan?
- What is the Max plan?
- What is the Team plan?
- Claude models overview
- Choosing the right model
- Claude Fable models on your plan
- Change the model, effort and thinking settings
- How do usage and length limits work?
- How large is the context window on paid Claude plans?
- Using Claude Code with your Pro or Max plan
- Use research on Claude
- Manage usage credits for paid Claude plans
- What is the Ultra plan?
- What's new in Claude Fable 5.2
- Claude Fable 5.2 launch offers