Claude Code can be paid for in two different ways: a flat monthly subscription with usage limits, or pay-per-token API billing with no ceiling except your budget. Which is cheaper depends on how many days a month you actually code with it and how many tokens each day uses. All prices below are as of 2026-11-19 and come from Anthropic's own pages, so check them before you commit.
The two billing models
Subscription. Anthropic's pricing page lists Pro at $20 per month billed monthly, or $17 per month billed annually ($200 up front). Max is $100 per month with 5x Pro usage, and Ultra is $200 per month with 10x Pro usage each week. Pro and Max have limits every five hours and every week; Anthropic's support article says verified Ultra accounts have no five-hour limit. Claude and Claude Code draw from the same pool, so chatting in the app uses the same allowance as coding.
API. You pay per million tokens (MTok). The Claude Developer Platform pricing page lists, for example, Claude Sonnet 5.5 at $2 input and $10 output, Opus 5.5 at $4 and $20, and Fable 5.2 at $8 and $40. Cache hits are far cheaper than fresh input: $0.10 per MTok on Sonnet 5.5 against $2. Anthropic's costs page says Claude Code charges by API token consumption.
Which one Claude Code uses
This is where people get surprised. According to the support article, Claude Code uses your plan allocation when you log in with Pro, Max or Ultra credentials. But if an ANTHROPIC_API_KEY environment variable is set, it uses that key instead, and you pay standard API rates. Two more details matter:
- The monthly API credits bundled with Max ($100) and Ultra ($200) cannot be used in Claude Code or the Claude apps.
- When you hit a limit, the options Anthropic lists are upgrading, enabling usage credits, using Console API credits, or waiting for the reset.
A break-even method
Anthropic's cost page reports that across enterprise deployments the average cost is around $13 per developer per active day, $150 to $250 per developer per month, and below $30 per active day for 90% of users. Dividing the plan prices by $13 gives a rough break-even in active days per month (my arithmetic):
| Plan | Monthly price | Days of average API spend it equals |
|---|---|---|
| Pro | $20 | about 1.5 |
| Max | $100 | about 7.7 |
| Ultra | $200 | about 15.4 |
Read this carefully. The $13 per active day figure is Anthropic's own average across enterprise deployments, and it may not reflect the model, the cache mix or the usage pattern in the worked example above, so treat it as a rough benchmark and not a measurement of your own spend. It is a benchmark for API-billed use at list prices, and the pages I read do not state a dollar value for what a plan's usage limits allow. So the table tells you how many days of average API spend a plan costs, not how much work a plan lets you do. If you code with Claude Code most weekdays at the average rate, API spend of roughly $150 to $250 would exceed the Max price, which is the situation where a subscription is most likely to save money, provided you stay inside its limits.
Measure your own tokens
Claude Code shows a session figure under /usage. Anthropic's example session on Sonnet 4.6 used 1.2k input, 5.3k output, 940.0k cache-read and 50.0k cache-write tokens and cost $0.55. At Sonnet 4.6 list prices ($3 input, $15 output, $0.30 cache read, $3.75 five-minute cache write per MTok) the arithmetic gives $0.5526, so the figure checks out. Charge the same tokens at Sonnet 5.5 prices ($2, $10, $0.10, $2.50) and the total is about $0.27, or about $0.35 if the cache writes are billed at the one-hour rate of $4.
Two lessons follow. Cached context is the bulk of the volume, so long sessions cost more than the visible prompt suggests, and choosing the model matters as much as choosing the plan. Anthropic's costs page also notes that for subscribers the session cost figure is not relevant for billing, since usage is included in the plan.
A practical decision rule
- Very light use, one or two active days a month: at the $13-per-day benchmark, API spend is about $13 to $26, close to Pro's $20, so it is nearly a toss-up. Past about 1.5 active days, Pro is the cheaper way to pay, as long as you stay inside its limits.
- Daily, heavy use: estimate monthly API-equivalent spend from
/usagefor two weeks, then compare against $100 (Max) and $200 (Ultra). - Automation, CI or team billing: use Console API billing, where spend limits and per-user reports exist.
- Mixed: stay on a plan and turn on usage credits so a limit does not stop work.
Take this further with Oakgen Agent
Agent can fetch the current Anthropic pricing pages, run the break-even calculations in a Python sandbox from numbers you give it, and export the comparison as an xlsx or md file.
Open this task in Oakgen Agent
The prompt is filled in for you. Nothing is sent until you press send.
Frequently asked questions
Is Claude Code included in the Claude Pro plan?
As of 2026-11-19, Anthropic's pricing page lists Claude Code as included on Pro, Max, Ultra, Team and Enterprise and not on the Free plan. Pro is $20 per month billed monthly or $17 per month billed annually ($200 up front), with usage limits that apply every 5 hours and every week.
How much does Claude Code cost on the API?
Anthropic's Claude Code cost page says Claude Code charges by API token consumption and that costs vary widely. Across enterprise deployments the average is around $13 per developer per active day and $150 to $250 per developer per month, with costs below $30 per active day for 90% of users. Treat these as benchmarks from large deployments, not a forecast for one person.
Can I use the API credits included with Max in Claude Code?
No. Anthropic's support article on using Claude Code with Pro, Max or Ultra says the monthly API credits included with Max and Ultra plans cannot be used in Claude Code or the Claude apps. To pay per token for Claude Code you would use a Console account with API credits or usage credits instead.
What happens if ANTHROPIC_API_KEY is set on my machine?
Anthropic's support article says that if an ANTHROPIC_API_KEY environment variable is set, Claude Code uses that API key instead of your Claude subscription, which results in API charges at standard API rates. To stay on your plan, log in with your plan credentials and decline the API credit option when it appears.


