AI token cost calculator
Estimate your monthly API bill on any set of models. Enter your traffic, the share of input you can cache, and the share you can send as batch jobs. For what a single token costs, see AI token cost.
Your traffic
A request is one API call. Accepts 100K, 2.5M and so on.
Prompt, instructions, documents and conversation history.
Repeated prefixes such as system prompts. Models without a cache rate are billed at the normal input price.
Batch jobs return within hours at a discount, where offered.
Count tokens in a piece of text
About 0 tokens. English averages about 4 characters per token; exact counts differ by model.
Open the full token counter, with exact OpenAI counts
Input per month
200M
tokens
Output per month
40M
tokens
Cheapest selected
$40.00
Claude Haiku 5.5 and GPT-6 Luna
Monthly cost at list prices
10 models · ·
$40.00 · $0.40 per 1K requestsTied lowest
$40.00 · $0.40 per 1K requestsTied lowest
$108 · $1.08 per 1K requests
$300 · $3.00 per 1K requests
$640 · $6.40 per 1K requests
$800 · $8.00 per 1K requests
$800 · $8.00 per 1K requests
$880 · $8.80 per 1K requests
$1,600 · $16.00 per 1K requests
$4,000 · $40.00 per 1K requests
The link keeps your traffic and models, so you can bookmark or share this estimate. Need a side-by-side table? Compare up to four of these.