The short answer: Claude Code has no price of its own. You reach it three ways: a Pro subscription ($20 a month, $17 billed annually), a Max subscription ($100 for 5x Pro usage, $200 for 20x), or pay-per-token API billing. Pro and Max share one usage allowance across Claude chat and Claude Code, so a busy coding day can use up your chat allowance too. Anthropic’s own docs put the average API cost at about $13 per developer per active day, so a full-time daily user is roughly a Max-versus-API decision, while a part-time user fits Pro.
The three routes side by side
Prices from claude.com/pricing and Anthropic’s docs, checked 29 September 2026 (the Max tier prices are from Anthropic’s Max help page, checked 28 September; the pricing page itself shows Max as “from $100”).
| Route | Price | How limits work | At the limit |
|---|---|---|---|
| Pro | $20 monthly, $17 per month billed annually | Five-hour session limit plus a weekly limit, shared with Claude chat | Wait for the reset, upgrade to Max, turn on usage credits, or use API billing |
| Max 5x | $100 a month | 5x Pro usage per session | Same options |
| Max 20x | $200 a month | 20x Pro usage per session | Same options |
| Team standard seat | $25 monthly, $20 per seat billed annually | More usage than Pro; includes Claude Code | Usage credits and admin spend limits |
| API (pay per token) | Opus 5.5 $4 in / $20 out per million tokens; Sonnet 5.5 $2 / $10; Haiku 4.5 $1 / $5 | No allowance: every token billed | Nothing to hit, the bill grows instead; set a workspace spend limit |
Anthropic does not publish how many tokens or messages a Pro or Max allowance holds, only the multiples. Any article giving you an exact message count for a plan is estimating.
Usage credits: the bridge between the two
Pro and Max subscribers can turn on prepaid usage credits. When the included limit runs out, further use bills at standard API rates and covers Claude Code as well as chat. You set a monthly spend cap or leave it unlimited. This is what makes the choice less risky than it looks: Pro plus credits behaves like a cheap entry price with an API-priced overflow.
One trap in Anthropic’s docs: if an ANTHROPIC_API_KEY environment variable is set, Claude Code uses it instead of your subscription, and you are billed per token. Check that variable if your subscription seems not to be draining.
A worked API example you can reproduce
Take one agent step that reads 50,000 tokens of context and writes 5,000 tokens, on Sonnet 5.5 at $2 and $10 per million:
- Input: 50,000 at $2 per million = $0.10
- Output: 5,000 at $10 per million = $0.05
- About $0.15 per step, before prompt caching (cache reads on Sonnet 5.5 are $0.20 per million, a tenth of the input rate, so long sessions cost less than this)
That is our arithmetic on published prices, not a measurement of a real session. Anthropic’s cost docs give a real-usage anchor instead: about $13 per developer per active day and $150 to $250 per developer per month across enterprise deployments, with 90% of users under $30 a day. At $150 to $250 a month, the $100 and $200 Max plans sit right where API billing would land for a heavy user, which is the case for Max.
Which route for whom
| You are… | Start with | Why |
|---|---|---|
| Trying Claude Code on a side project | Pro | $20, and usage credits as a safety valve |
| Using it most working days on real code | Max 5x, then check /usage | Flat price; compare against Anthropic’s $150 to $250 a month API average |
| Running long autonomous sessions or several agents at once | Max 20x, or API with a spend limit | Usage climbs fast with subagents and long context |
| A team with an admin | Team seats plus usage credits | Central billing and per-member limits |
| Building a product on Claude models | API | Not a coding-assistant question |
Anthropic’s cost page lists the habits that cut usage: clear context between unrelated tasks, use Sonnet rather than Opus for routine work, and keep CLAUDE.md short. Our CLAUDE.md guide and context window guide cover the two most practical. Compared with editors: Claude Code vs Cursor sets the prices side by side, and Cursor pricing explains the other pool model. The tool itself is in our Claude Code review. Prices for all twelve tools sit in the pricing index.
Once you have picked a route, how to use Claude Code walks through install and the first task.
Method
Researched from Anthropic’s pricing page, help centre and docs on 28 to 29 September 2026. We have not run these plans to their limits; allowance sizes are not published and are not estimated here. The dollar-per-day figure is Anthropic’s, from enterprise deployments, and may not match a solo developer. Re-checked every 30 days.
Common questions
Is Claude Code free?
No. It requires a paid Claude plan (Pro, Max, Team or Enterprise) or an API account billed per token.
Is Claude Max worth it over Pro?
If you keep hitting Pro’s five-hour or weekly limit, Max 5x removes that for $80 more a month. If you rarely hit it, Pro plus usage credits is cheaper. Check the usage bars in Settings before upgrading.
Does Claude Code use my chat allowance?
Yes. On Pro and Max, activity in Claude and Claude Code counts against the same limits, per Anthropic’s help centre.