Claude Code pricing, October 2026: Pro, Max and Team plans vs paying per token

Claude Code pricing comes in two shapes. A Claude plan includes Claude Code from $17 a month and caps how much you use. An API key has no plan price and bills every token. The plans below are Anthropic's; the last two rows are the per-token options.

A plan's usage resets on a rolling five-hour window, with a weekly limit on top, and Claude Code shares that pool with Claude chat. Anthropic does not publish the allowance in tokens ("there's no fixed message count", its pricing FAQ). So the useful comparison is your month's token bill against the plan price, which the next tables work out.

The same monthly money spent per token, counted in Claude Code turns of 80,000 cached, 2,000 new input and 1,500 output tokens. That is one turn late in a working session, explained under the three prices below.

On Sonnet 5.5, Pro's $20 buys 740 turns at Anthropic's list price and 1,550 here. At 150 turns a working day that is 10.3 days of steady use on a key here, with nothing resetting on a five-hour clock.

A Claude Code bill on an API key comes from three rates: input, cached input and output, each per million tokens. Claude Code sends the whole conversation again on every request, so most of the tokens in a session are the same tokens, read back from cache. The cached rate decides most of the bill.

Claude Opus 5.5 is $2.10 in and $6.00 out here, against $4.00 and $20.00 at Anthropic: 47% under list on input and 70% on output. The other rows read the same way. Every Claude model on the key gets the same test prompt once a day, with the token counts public on the real-model check.

Take one turn late in a session: 80,000 tokens read from cache, 2,000 new input tokens and 1,500 output tokens. Cache writes are left out.

The arithmetic for Sonnet 5.5 at list: 80,000 × $0.10 + 2,000 × $2.00 + 1,500 × $10.00, all per million, is $0.027. The same turn here is $0.0129.

On an API key the prompt cache lasts five minutes by default; on a subscription it lasts an hour (Anthropic's Claude Code cost docs, read 2026-10-18). Step away for longer and the next request pays the input rate for the whole context: 82,000 input tokens plus 1,500 output.

That one turn costs about five times the cached one. A long break in the middle of a session costs one of these; a session that keeps moving pays the cached rate.

Three sizes of day, each made of the turn above with every turn hitting the cache. The sizes are assumptions, not measurements; change the turn count and the table scales with it.

For scale, Anthropic's own figure across its enterprise deployments is about $13 per developer per active day at list price, and under $30 a day for 90% of users (the same docs, read 2026-10-18). That is context, not a forecast for your sessions.

The same three days over 20 working days. This is the number to hold against a plan price.

A steady month on Opus 5.5 is $162.00 at list and $68.40 here. Anthropic's own average is $150-250 per developer per month at list (the cost docs). On Sonnet 5.5 the same month is $81.00 at list and $38.80 here.

A plan costs the same whatever you use. It is the cheaper route when your month on a key would cost more than the plan and the five-hour window rarely stops you.

A key costs what you use. It is the cheaper route when your month comes in under the plan price, and when you code in bursts: a quiet week costs nothing, and a long day never waits for a reset. Each key can carry its own monthly spend limit, so one developer or one CI job has a ceiling of its own.

Two details from Anthropic's Help Center (Claude Code with Pro or Max, read 2026-10-18): the monthly API credit that comes with Max can't be spent in Claude Code, and an ANTHROPIC_API_KEY variable in your shell makes Claude Code bill that key instead of your plan.

On an API key, Claude Code computes the session cost on your machine from its token counts at Anthropic's list price (the cost docs). It is an estimate at list, whatever host you point it at. On a plan, /usage shows plan usage bars instead. The real bill for a key here is in the console, which records what each request was charged.

Thinking tokens bill as output and cannot be switched off on Opus 5.5, Sonnet 5.5, Haiku 5.5 or the Fable models (the cost docs). A lower effort level produces fewer of them.

Two variables. The Claude Code guide has the settings.json form and the model variables, and how to get a Claude API key covers Anthropic's own key.

export ANTHROPIC_BASE_URL="https://api.uplinkswitch.com/v1"
export ANTHROPIC_AUTH_TOKEN="sk-uplink-live-..."
claude

Prompt caching works through the gateway; cached input bills at the cached rate in the price table. To go back to your plan, unset both variables and run /login with your Claude account.

• /clear between tasks, so the next request does not carry the last task's context.
• The right model per task. Haiku 5.5 for subagents and small edits.
• A shorter CLAUDE.md. It is in every request.
• A lower effort level where the task allows it.
• Keep sessions moving. A turn after the five-minute cache expiry pays the input rate for the whole context.
• A monthly spend limit on the key, so a runaway loop stops at a number you chose.

How much does Claude Code cost? $17 to $200 a month on a personal plan, $20 to $125 a seat on Team, or per token on an API key. On a key, a steady month of 150 turns a working day is $81.00 on Sonnet 5.5 at list and $38.80 here.

Is the $20 Claude Code plan worth it? It is when your month on a key would cost more than $20 and the Pro limits rarely stop you. $20 buys 740 Sonnet 5.5 turns at list, or 1,550 here. Count your turns for a week and multiply.

Is Claude Code free? No. The Free plan does not include Claude Code. It needs a paid plan or an API key with credit on it.

Is it cheaper to use Claude Code with the API or a subscription? Cheaper on a key when your month in the table above comes in under the plan price; cheaper on the plan when it comes in over and you stay inside the limits.

Can I spend my Max plan's API credits in Claude Code? No. Anthropic's Help Center says the monthly API credits on Max and Team can't be used in Claude Code or the Claude apps.

Why is Claude Code expensive? Every request carries the whole conversation. At 80,000 tokens of context, even the cached rate adds up over hundreds of turns, and any turn after a break pays the input rate for all of it.

Where do I check my own numbers? Export your Anthropic usage and re-price it on the savings page. Every Claude model's price is on the Claude pricing hub, and one request at a time is worked in Claude API cost per request.