| Item | Price per million tokens |
|---|---|
| Input | $4 |
| Output (incl. thinking) | $20 |
| Cache write, 5 minutes | $5 |
| Cache write, 1 hour | $8 |
| Cache read | $0.2 |
| Batch API input / output | $2 / $10 |
Model id: claude-opus-5-5
What it costs in practice
| Request | One request | 1,000 requests | One request, Batch API |
|---|---|---|---|
| Short chat reply 2,000 in / 300 out | $0.014 | $14.00 | $0.0070 |
| Agent turn with tools and history 12,000 in / 800 out | $0.064 | $64.00 | $0.032 |
| Summarize a 40-page document 30,000 in / 1,000 out | $0.140 | $140.00 | $0.070 |
Prompt caching example
An agent sends the same 10,000-token system prompt and tool list on every call, plus 1,000 new tokens, and gets 500 tokens back. Over 20 calls inside a 5-minute cache window:
- Without caching: $1.08
- With caching (one cache write, then 19 cache reads): $0.368, about 66% less
Compared with other Claude models
- Claude Fable 5.2: $8 / $40, about 2× the price of Opus 5.5 for the same tokens.
- Claude Sonnet 5.5: $2 / $10, about 2× cheaper than Opus 5.5 for the same tokens.
- Claude Haiku 5.5: $0.1 / $0.5, about 40× cheaper than Opus 5.5 for the same tokens.
Side-by-side pages with worked examples:
Input + output price per million tokens, compared as a simple sum. Your real ratio depends on how many input versus output tokens your workload uses; the calculator compares models on your own numbers.
What a typical app costs on Claude Opus 5.5: chatbot, support bot, coding agent, document q&a (rag), document summarization, classification & extraction.
Questions
How much does Claude Opus 5.5 cost?
$4 per million input tokens and $20 per million output tokens. Prompt cache reads cost $0.2 per million tokens; cache writes cost $5 (5-minute) or $8 (1-hour) per million tokens.
What is the Claude Opus 5.5 model id?
Use "claude-opus-5-5" as the model parameter in the Messages API.
Is there a cheaper way to run Claude Opus 5.5?
Yes. Batch API requests cost 50% of the normal price, and prompt caching cuts the cost of repeated context. In the caching example on this page, 20 calls that share a 10,000-token prefix cost about 66% less with caching.