Home / Guides / Claude pay-as-you-go pricing
Claude pay-as-you-go pricing
You can pay Claude by the token in three ways, and all three use Anthropic's standard API rates. On Pro or Max, usage credits keep Claude and Claude Code running after you hit your 5-hour or weekly limit, billed on top of the subscription. With an API key from a Claude Console account you skip the subscription and pay for every token. On Enterprise, each seat costs $20 a month plus usage at API rates. Per million tokens, Opus 5.5 costs $4 of input and $20 of output, Sonnet 5.5 $2 and $10, Haiku 5.5 $0.10 and $0.50 for prompts up to 100,000 tokens, and Fable 5.1 $10 and $50.
Whether that beats a plan depends on how often you code. Anthropic puts the average Claude Code bill at about $13 per developer per active day, which passes Pro's $20 on your second active day and Max 5x's $100 after about eight. Input read from the prompt cache costs a twentieth of fresh input on Opus 5.5 and Sonnet 5.5, and in Anthropic's own Claude Code example 94% of the tokens are cache reads, so the headline input price says less about your bill than it seems.
Checked 29 October 2026 against claude.com/pricing, Anthropic's usage credits article and the Claude Code cost docs · By Soren Starck
| Model | Input | Output | Cache write | Cache read |
|---|---|---|---|---|
| Fable 5.1 | $10 | $50 | $12.50 | $0.25 |
| Opus 5.5 | $4 | $20 | $5 | $0.20 |
| Sonnet 5.5 | $2 | $10 | $2.50 | $0.10 |
| Haiku 5.5, prompts up to 100K tokens | $0.10 | $0.50 | $0.125 | $0.01 |
| Haiku 5.5, prompts over 100K tokens | $0.50 | $2.50 | $0.625 | $0.05 |
US dollars per million tokens, Claude API list prices from claude.com/pricing. Cache writes shown are the 5-minute kind; a 1-hour cache write costs twice the input price. Batch processing halves input and output, US-only inference adds 10%, and fast mode on Opus 5.5 doubles the price. Usage credits on Pro and Max bill at these same rates.
SessionWatcherSee your Claude limits before
you start paying per token.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
On Windows? Claude usage in the Windows taskbar: the same limits and alerts on Windows 10 and 11.
How do usage credits work on Claude Pro and Max?
They take over where your plan stops. With usage credits on, a request that would hit your 5-hour session limit, your weekly limit or, on Max, the separate Fable weekly limit goes through anyway and is billed at the standard API rate. The limit notice changes to "Now using credits", and your limits keep resetting on their usual schedule. Credits cover Claude chats and Claude Code alike, because both draw from the same plan.
To set them up, open Settings > Usage on claude.ai, find Usage credits and click Enable. You add a payment method, set a monthly spend limit (or choose unlimited), then prepay with Add funds; auto-reload can top the balance up when it drops below a threshold you pick. Spending stops at your monthly limit, or when the balance runs out with auto-reload off. If you subscribed in the iOS or Android app, you can only turn credits on in the web version of Claude.
How much does Claude extra usage cost?
The API rates in the table above, with no separate price list. Anthropic sets no price per message, because a message costs whatever tokens it uses: a long conversation, attached files and a larger model all cost more. Credits are charged separately from your subscription and show up as extra charges on your bill. On Pro, Fable is only available through usage credits; Max 5x and Max 20x include Fable within half of their weekly limits, and credits pay for anything beyond.
Can you use Claude Code pay as you go?
Yes, in either of two ways. Turn on usage credits on your Pro or Max plan and Claude Code keeps working past your limits. Or sign Claude Code in with a Claude Console account instead of your subscription, and every token is billed at API rates with no plan at all; Anthropic's pricing FAQ names it as the option for heavy coding sessions. With a Console account, the Session block of /usage shows what the session has cost so far. Claude Code works that figure out on your machine from token counts at list price, so it is an estimate, and the Usage page in the Claude Console is the bill.
Is pay-as-you-go cheaper than Pro or Max?
Only if you code with Claude a few days a month. Anthropic's Claude Code cost docs put the average across enterprise deployments at about $13 per developer per active day and $150 to $250 per developer per month, with 90% of users under $30 per active day. At $13 a day, the bill reaches each plan's price after:
| Plan | Price a month | Pay-as-you-go passes it after |
|---|---|---|
| Pro | $20 | 2 active days |
| Max 5x | $100 | 8 active days |
| Max 20x | $200 | 16 active days |
Arithmetic on Anthropic's average of $13 per active day; your own daily cost can be far above or below it.
If you code most working days, the average API bill is more than Max 5x and close to Max 20x, so a plan costs less for as long as its limits hold you, and usage credits cover the weeks they don't. What each plan includes is in Claude Code pricing and Claude Pro vs Max; the subscription on its own, monthly against yearly, is in Claude Pro price.
What does a Claude Code session cost per token?
Anthropic's cost docs show one session from /usage: 1,200 input tokens, 5,300 output tokens, 940,000 cache reads and 50,000 cache writes, $0.55 on Sonnet 4.6. Cache reads are 94% of those tokens. The same session at today's rates (input, output, cache read and cache write, added up):
| Model | Session cost | Input + output + cache read + cache write |
|---|---|---|
| Sonnet 5.5 | $0.27 | $0.002 + $0.053 + $0.094 + $0.125 |
| Opus 5.5 | $0.55 | $0.005 + $0.106 + $0.188 + $0.25 |
| Fable 5.1 | $1.14 | $0.012 + $0.265 + $0.235 + $0.625 |
The output tokens and the cache writes make most of the bill, and the 1,200 tokens of fresh input hardly count. Each time the cache has to be written again (a new session, or one left idle past the cache lifetime), those tokens cost the write price instead of the read price: 25 times more per token on Sonnet 5.5 for a 5-minute cache, 40 times for a 1-hour one.
To run the same sum on your own work, SessionWatcher reads the tokens Claude Code logs on your Mac and estimates what they would cost at API rates, model by model, for today, the last 7 days, this month and the last 30 days, next to your live 5-hour and weekly limits. A week of it tells you whether a plan, usage credits or an API key costs you least, with your own numbers instead of an average.
Do Max's monthly API credits pay for usage credits?
No. Max 5x comes with $100 and Max 20x with $200 of API credits a month for building on the Claude Platform, and Team plans with up to $500 a month in pooled API credits. Anthropic says monthly API credits do not apply to usage credits in Claude, Claude Code or Claude Cowork, so going past a Max limit is charged on its own.
How to keep pay-as-you-go spending under control
- Set a monthly spend limit on usage credits, and leave auto-reload off until you know your monthly figure.
- Settings > Usage shows your month-to-date credit spending and past use; Anthropic also sends alerts as you near your spend limit.
- Use Sonnet or Haiku for routine work, as Anthropic itself advises: Opus 5.5 costs twice Sonnet 5.5 per token, and Fable 5.1 five times.
- Know where your plan limits stand before you cross into credits. Settings > Usage and
/usageshow them when you look; SessionWatcher keeps them in the menu bar and alerts you before you reach them. Every way to check your limits and the credits you have left is in how to check Claude Code usage.
SessionWatcherKnow when you are about to pay extra
for Claude, before it happens.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
Frequently asked questions
Is Claude pay as you go?
It can be. The Claude API bills per token with no subscription, and on Pro, Max 5x and Max 20x you can turn on usage credits, which keep Claude and Claude Code working past your plan limits at standard API rates. The Free plan has no usage credits.
How much do Claude usage credits cost?
The standard API rate of the model you use, per million tokens: Opus 5.5 $4 input and $20 output, Sonnet 5.5 $2 and $10, Haiku 5.5 $0.10 and $0.50 for prompts up to 100,000 tokens, Fable 5.1 $10 and $50. Anthropic sets no price per message, because a message costs whatever tokens it uses.
Does Claude Pro charge per token?
Not for the usage the plan includes: $20 a month, or $17 a month billed yearly, covers your 5-hour session and weekly limits whatever the token count. Tokens are charged only once you turn on usage credits and go past a limit, and those charges come on top of the subscription.
Can I turn Claude usage credits off?
Yes, at any time in Settings > Usage. With them off you only have your plan’s included usage, and Claude stops at your limit until it resets.
Is Claude Enterprise pay as you go?
Partly. Enterprise costs $20 per seat per month, billed annually with a 20-seat minimum, plus usage billed at API rates, so the bill grows with the models and tasks your team runs.
Sources
All read on 29 October 2026:
- claude.com/pricing: API rates per model, batch, US-only inference and fast mode, the plan table (usage credits, Fable, monthly API credits), Enterprise pricing, the usage FAQ.
- Claude API pricing docs: 1-hour cache writes and the cache read multipliers.
- Manage usage credits for paid Claude plans: how credits switch on, billing, setup, spend limits and auto-reload.
- Claude Code: manage costs effectively: the average cost per developer, the /usage cost figure and the sample session.