As of November 11, 2026: Anthropic lists Claude Haiku 5.5 API pricing starting at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens. Prompts above that threshold use higher listed rates. These are Claude Platform API prices, not a Claude subscription fee.
Claude Haiku 5.5 API rates
Anthropic identifies the model as claude-haiku-5-5. The following Claude Platform rates are per million tokens (MTok):
| Token type | Prompts up to 100,000 tokens | Prompts over 100,000 tokens |
|---|---|---|
| Input | $0.10 | $0.50 |
| Output | $0.50 | $2.50 |
| Five-minute cache write | $0.125 | $0.625 |
| One-hour cache write | $0.20 | $1.00 |
| Cache read | $0.01 | $0.05 |
Anthropic also lists a 50% discount on input and output for the Batch API. Check the official pricing page for eligibility, the exact threshold calculation, cache rules, and any other applicable charges before budgeting. Other providers and cloud platforms may have different pricing.
For comparison, the Haiku 4.5 model page lists $1 per million input tokens and $5 per million output tokens. Confirm the exact model and provider before applying either rate.
Example: estimate a short-prompt workload
Suppose an application uses Haiku 5.5 on the Claude Platform, with each prompt at or below 100,000 tokens. Across a billing period it sends 100 million input tokens and receives 20 million output tokens, without caching or Batch API discounts:
- Input: 100 × $0.10 = $10.
- Output: 20 × $0.50 = $10.
- Estimated token total: $20.
This is an illustration, not a quote. Actual invoices depend on metered tokens, the tier applied to each request, caching, batch eligibility, and any additional charges. Check the final current billing terms with the provider.
What are Haiku 5.5's main limits?
Anthropic's Haiku 5.5 model page lists a 1 million-token context window, up to 128,000 output tokens for the synchronous Messages API, text and image input, and text output. It also lists adaptive thinking with a medium default effort. The Batch API has a separately documented higher beta output limit with an applicable header.
For tool calling, structured outputs, rate limits, and account-specific availability, consult the current documentation for your API channel before relying on those features. Avoid substituting another gateway's model limits for the direct Claude Platform specifications.
FAQ
Is Claude Haiku free?
The rates above describe paid API token usage. A subscription, trial, or another provider's offer has separate terms; verify the relevant account and service.
Does a long prompt cost more?
The published Haiku 5.5 table lists a higher per-token tier for prompts over 100,000 tokens. Confirm exactly how Anthropic measures the threshold and applies it to your request before estimating a long-context workload.
Editorial status: manually prepared; pending human review. Not published. Model availability, limits, and prices must be rechecked before publication.
Sources
https://platform.claude.com/docs/en/about-claude/pricing
https://platform.claude.com/docs/en/models/haiku-4-5/overview
https://platform.claude.com/docs/en/models/haiku-5-5/overview