Claude Pro costs $20 a month and includes Claude Code. Max 5x ($100) and Max 20x ($200) raise your usage limits and each include API credits equal to their price ($100 and $200 per month) for the Claude Platform. At list price, $20 of Sonnet 5.5 API spend buys about 1,577 chat answers or 251 code reviews, which is a useful yardstick for deciding which plan fits.
What each plan includes
My data lists these plans as of 2026-10-11:
- Claude Free ($0): Chat on web, desktop and mobile, web search, file creation and code execution. No Opus or Fable access and no Claude Code.
- Claude Pro ($20): More usage than Free and Claude Code included. Fable is available only through usage credits.
- Claude Max 5x ($100): 5x the usage of Pro, $100 per month in API credits, Fable limited to 50% of weekly limits, and priority access at high traffic times.
- Claude Max 20x ($200): 20x the usage of Pro, $200 per month in API credits, the same Fable limit and priority access.
My data does not convert plan usage limits into tokens or messages, so I cannot say how many chats a Pro or Max plan covers. What I can compare is API-equivalent value: how many tasks a given dollar amount buys at list price.
What the API charges
Sonnet 5.5 lists at $2.00 per 1M input tokens, $10.00 per 1M output tokens and $0.10 per 1M cached input tokens. Opus 5.5 lists at $4.00, $20.00 and $0.20. Both have a 1,000,000-token context window and both use reasoning always. Reasoning tokens are billed as output tokens.
Per-task costs for the two models:
| Task | Sonnet 5.5 per task | Sonnet 5.5 per 1,000 | Opus 5.5 per task | Opus 5.5 per 1,000 | Basis |
|---|---|---|---|---|---|
| Answer a chat question | 1.3¢ | $12.68 | 2.5¢ | $25.36 | Sonnet measured, Opus estimated |
| Review 500 lines of code | 8.0¢ | $79.66 | 16¢ | $159.33 | Sonnet measured, Opus estimated |
| Run a 50-step coding agent | 84¢ | $841.64 | $1.68 | $1,683.28 | Modeled |
The median of all 326 models is 0.24¢ for a chat answer, 1.4¢ for a code review and 48¢ for the agent run. Sonnet 5.5 sits above the median on all three, and Opus 5.5 costs about twice as much as Sonnet on each. See the chatbot answer, code review and 50-step agent pages for the full rankings.
Two details matter for agent work:
- The agent scenario is an 8,000-token system prompt, 50 steps, 1,500 extra context tokens per step and 300 visible output tokens per step.
- Caching changes the bill a lot. Sonnet 5.5's agent run is $6.52 without caching against 84¢ with it. Opus 5.5's uncached run is $13.04.
Break-even counts
The table shows how many tasks a given monthly amount buys at list API prices. Pro lists no API credit, so its column shows what $20 would buy if spent on the API instead. Treat these as rough yardsticks, since plan usage is not priced per token.
| Plan | Monthly price | API credit included | Sonnet 5.5 chats | Opus 5.5 chats | Sonnet 5.5 code reviews | Opus 5.5 code reviews |
|---|---|---|---|---|---|---|
| Pro | $20 | None listed | 1,577 | 788 | 251 | 125 |
| Max 5x | $100 | $100 | 7,886 | 3,943 | 1,255 | 627 |
| Max 20x | $200 | $200 | 15,772 | 7,886 | 2,510 | 1,255 |
How to read it:
- Pro: If you would otherwise pay the API directly, roughly 1,500 Sonnet chats a month costs about what Pro does. Below that, pay-as-you-go is cheaper on tokens alone, but Pro adds Claude Code and app access.
- Max 5x and 20x: The credits equal the plan price, so the subscription matches API spend of the same size and adds higher usage limits and priority access on top.
For agent runs the counts are much smaller:
- At 84¢ per Sonnet 5.5 run, $20 pays for about 23 runs. At $1.68 per Opus 5.5 run, it pays for about 11.
- The Max 5x credit covers about 118 Sonnet runs, and the Max 20x credit about 237.
- With no caching, a $6.52 Sonnet run cuts $20 to roughly 3 runs.
Which option fits
- Stay on Free for occasional chat. It has no Opus, no Fable and no Claude Code.
- Pick Pro if you mainly want Claude in the app plus Claude Code. Your chat volume would have to exceed roughly 1,500 Sonnet answers a month before API pricing matched the plan.
- Pick Max 5x or 20x if you already spend $100 or $200 a month on the API, or need the higher usage limits. The credits offset that spend.
- Use the API directly for production pipelines that need predictable per-task cost and control of caching and output length. My calculator lets you plug in your own volumes.
Check your own workload on the Sonnet 5.5 and Opus 5.5 model pages, and see my methodology for how I measure. Plans change, so see my plans page for the latest.
FAQ
Is Claude Max cheaper than paying for the API?
It matches API spend at the same size. Max 5x includes $100 of API credits for $100 a month, so the credits offset the price and you also get higher usage limits. If you would spend less than that on tokens, pay-as-you-go is cheaper.
How many chats does $20 of API spend buy?
At list price, about 1,577 Sonnet 5.5 chat answers or about 788 Opus 5.5 answers, based on my measured Sonnet and estimated Opus per-task costs. Your prompts and output length will change this.
Does caching change the comparison?
Yes, for agent work. Sonnet 5.5's 50-step run costs 84¢ with caching and $6.52 without it in my model. If your setup cannot reuse cached input, the API gets much more expensive relative to a flat plan.