How does Anthropic typically price Claude API usage?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Anthropic charges per million tokens — and output tokens typically cost more than input tokens, since generating is harder than reading.
Full explanation below image
Full Explanation
Anthropic's API pricing is token-based, with distinct per-token rates for input (prompt) tokens and output (completion) tokens. Output tokens are typically priced higher than input tokens. Prices vary by model — Haiku is cheapest per token, Opus is most expensive. This incentivizes using larger models only when needed and optimizing prompt lengths. Option A (subscription) doesn't match Anthropic's pay-per-use pricing model for API access. Option C (per call) would disadvantage short requests. Option D (by response time) is not how Anthropic prices its API.