Claude token counter & Anthropic cost calculator
Paste any text to see how many tokens Claude reads it as, and what it costs on each Anthropic model. Counts update as you type.
| Model | Input $/M | Output $/M | Context | Per call | Per month |
|---|---|---|---|---|---|
|
Claude Fable 5
Anthropic |
$10.00 | $50.00 | 200K | — | — |
|
Claude Opus 5
Anthropic |
$5.00 | $25.00 | 200K | — | — |
|
Claude Sonnet 5
Anthropic |
$2.00 | $10.00 | 200K | — | — |
|
Claude Opus 4.8
Anthropic |
$5.00 | $25.00 | 200K | — | — |
|
Claude Sonnet 4.6
Anthropic |
$3.00 | $15.00 | 200K | — | — |
|
Claude Haiku 4.5
Anthropic |
$1.00 | $5.00 | 200K | — | — |
Rates are USD per million tokens, published list price, checked 2026-08-31. Batch and cached-input discounts are not applied. Confirm against the provider before committing a budget.
How Claude counts tokens
Claude uses its own byte-level BPE vocabulary, and it is a smaller one than OpenAI's. A smaller vocabulary merges less aggressively, so the same English sentence generally costs slightly more tokens on Claude than on GPT — this calculator models that gap at about ten percent.
The counter above is an estimate, not Anthropic's own merge table. It models the behaviour that actually drives the number — a leading space merging into the word after it, digits grouping, characters outside the Latin range costing several times more — which is close enough to size a prompt and budget a workload, and deliberately not close enough to reconcile an invoice against.
The same text is not the same token count here
This is the trap specific to Claude, and it catches people who sized their budget on an OpenAI count. If you measured a prompt at 10,000 tokens against GPT and moved it to Claude unchanged, expect roughly 11,000 — for identical text, with no change on your side. It matters twice over: once on the bill, and once at the context window, because a prompt that fitted with a little room to spare may not fit any more. The tokeniser tabs above the table show the gap on your own text rather than on an average.
Claude Sonnet 5: Introductory rate through 2026-08-31; $3.00 / $15.00 after.
What the 6 Claude models cost
Take one representative call — a 2,000-token prompt and a 600-token reply, which is roughly two pages in and one page out — and run it 100 times a day.
the cheapest here
the dearest here
for identical work
on the dearest, vs $15.00 on the cheapest
Anthropic's spread between its cheapest and dearest model is real but narrower than OpenAI's, so the model choice matters less than the prompt does. Where Claude pays back most is prompt caching: a long system prompt or document reused across many calls can be cached rather than resent, which changes the arithmetic for exactly the workloads that look most expensive in the table. The rates here are list price with no caching applied.
Which Claude model to use
Haiku for volume, Sonnet for almost everything, Opus for the jobs where being wrong is expensive. The Claude range is unusual in that the middle model is genuinely the default rather than a compromise — Sonnet handles long-document work and careful writing well enough that reaching for Opus is a deliberate decision about difficulty, not a reflex. If your workload is one long document and one careful answer, Sonnet is the model to price first.
Context windows
The Claude models here run from 200K tokens to 200K. That window is shared: your prompt, every earlier turn in the conversation, and the reply all have to fit inside it together. The table above marks a model in red when the text you have pasted plus the reply length you set would not fit — which is usually how people discover that a window is not as generous as the headline number suggests.
Questions
How many tokens does Claude count for 1,000 words?
About 1,400 for ordinary English prose — a little more than the same text on GPT, because Claude's vocabulary merges less aggressively. Code and non-Latin scripts run higher again. Paste your real text above for a figure that reflects it.
Is this the official Anthropic token count?
No, it is an estimate. Anthropic publishes a token-counting endpoint that returns the exact figure for a given model; this page models the general behaviour of the tokeniser instead, so it needs no API key and no signup. For prose it is close enough to budget with, and it is deliberately not close enough to bill with.
Does Claude's free chat use tokens?
The claude.ai chat interface is a subscription with usage limits, not a per-token bill. Tokens are billed through the API, which is what this page prices. The context window applies to both, so a document too long for the API is too long for the chat as well.
Why is the Sonnet price marked as introductory?
Because it is dated and will change. Anthropic set an introductory rate with a published end date, after which it rises; the table shows the note against that model. It is worth knowing before you build a budget on the current number rather than after.
Counting for a different provider
The same text costs a different number of tokens on every family, so if you are comparing providers, compare on your own text rather than on a rule of thumb.
Rates are USD per million tokens, published list price, checked 2026-08-31. Batch and cached-input discounts are not applied. Confirm against Anthropic before committing a budget.
More free tools
All free, all instant, none of them need an account.