API pricing · checked October 8, 2026
Claude Haiku 5.5 pricing
Claude Haiku 5.5 is the cheapest model in Anthropic's current lineup: $0.10 per million input tokens and $0.50 per million output tokens when your prompt is 100K tokens or less. Below: the full rate card, how the 100K-token rule works, what real workloads cost, and how to use Haiku 5.5 without paying anything.
- input
- $0.10
- output
- $0.50
per 1M tokens, for prompts up to 100K tokens
- Prompts over 100K tokens
- $0.50 / $2.50
- Cache reads
- $0.01
- Batch API
- 50% off
- Against Haiku 4.5
- up to 90% cheaper
The full rate card
Every Claude Haiku 5.5 price on the Claude API, in US dollars per million tokens. Which column applies depends on the size of each request's prompt.
| Per 1M tokens | Prompt ≤ 100K tokens | Prompt > 100K tokens |
|---|---|---|
| Input | $0.10 | $0.50 |
| Output | $0.50 | $2.50 |
| Cache write (5 minutes) | $0.125 | $0.625 |
| Cache write (1 hour) | $0.20 | $1 |
| Cache read | $0.01 | $0.05 |
| Batch input | $0.05 | $0.25 |
| Batch output | $0.25 | $1.25 |
Source: Anthropic's pricing documentation, checked October 8, 2026. The same rates apply on Claude Platform on AWS and Microsoft Foundry; Amazon Bedrock and Google Cloud publish their own prices.
How the 100K-token rule works
Haiku 5.5 is the only model in Anthropic's current lineup that is priced by prompt length. Each request is billed on one of two rate cards. If its prompt is 100,000 tokens or less, it uses the low card. If the prompt is longer, everything in that request — input, output and cache — is billed on the high card, which costs 5× more.
So the line matters. A 90K-token prompt with a 2,000-token answer costs about $0.010. Grow the prompt to 120K tokens and the same answer costs $0.065: 6.5× more for a third more input.
100K → $0.011
110K → $0.06
Prompt size (tokens)
100K tokens is roughly 55,000 English words. If your prompts drift past it, trim old chat turns, retrieve only the passages each request needs, or summarize history. And if a job truly has to read more than 100K tokens on every call, Haiku 5.5's high card is still 4× cheaper than Claude Sonnet 5.5.
Haiku 5.5 vs other Claude models
Per token, Haiku 5.5 is 10× cheaper than Haiku 4.5, 20× cheaper than Sonnet 5.5 and 40× cheaper than Opus 5.5.
| Model | Input | Output | Cache read | Batch (in / out) |
|---|---|---|---|---|
| Claude Haiku 5.5prompts ≤ 100K tokens | $0.10 | $0.50 | $0.01 | $0.05 / $0.25 |
| Claude Haiku 4.5 | $1 | $5 | $0.10 | $0.50 / $2.50 |
| Claude Sonnet 5.5 | $2 | $10 | $0.10 | $1 / $5 |
| Claude Opus 5.5 | $4 | $20 | $0.20 | $2 / $10 |
| Claude Fable 5.1 | $10 | $50 | $0.25 | $5 / $25 |
Token counts are held equal across models. Haiku 4.5 uses an older tokenizer that counts the same text as roughly 23% fewer tokens, so its real gap is a little smaller. Anthropic's own figure, with the tokenizer accounted for: Haiku 5.5 is 90% cheaper than Haiku 4.5 for requests up to 100K tokens and 50% cheaper above.
Haiku 5.5 cost calculator
Describe your workload to estimate a monthly bill at list price, and see what the same traffic would cost on other Claude models.
Start from an example
Count thinking tokens too: they are billed as output.
Batch API
Half price for asynchronous jobs
Claude Haiku 5.5
$45.00per month
$0.00045 per requestbilled on the ≤ 100K-token card
Same workload on other Claude models
- Haiku 5.5$45.00
- Haiku 4.5$450.0010× Haiku 5.5
- Sonnet 5.5$900.0020× Haiku 5.5
- Opus 5.5$1,80040× Haiku 5.5
- Fable 5.1$4,500100× Haiku 5.5
Estimates at Claude API list prices. Cache writes, tool definitions and web searches are not included.
Other costs to know
Token prices are most of the bill, but tools and deployment options add charges of their own.
| Item | Cost |
|---|---|
| Web search | $10 per 1,000 searches, plus the result tokens as input |
| Web fetch | No extra fee; fetched content is billed as input tokens |
| Code execution | Free alongside web search or web fetch; otherwise 1,550 free hours a month, then $0.05 per container-hour |
| Tool definitions | 286 extra input tokens per request with tools (406 when a tool is forced), plus your tool schemas |
| Thinking | Billed as output tokens; a lower effort level spends fewer |
| US-only inference | Setting inference_geo to us multiplies every token price by 1.1 |
| Amazon Bedrock and Google Cloud | Priced by the cloud provider; regional endpoints cost 10% more than global ones |
Five ways to cut your Haiku 5.5 bill
- 01
Stay under 100K tokens
The low rate card is 5× cheaper. Trim history and retrieve only what each request needs.
- 02
Cache your stable prefix
Cache reads cost $0.01 per million tokens, a tenth of normal input. System prompts, tool definitions and shared documents are ideal.
- 03
Batch what can wait
The Message Batches API halves every token price for asynchronous jobs.
- 04
Pick the effort level
Thinking tokens bill as output. Low effort suits chat and classification; medium is the default.
- 05
Re-measure after migrating
Haiku 5.5 counts the same text as about 30% more tokens than Haiku 4.5. Recount your prompts before you budget.
How to use Haiku 5.5 for free
On Haiku55
Chat with Claude Haiku 5.5 in your browser: 5 free messages a day without an account, then 30 free credits every day once you sign in, with web search and link reading included.
In the Claude apps
Anthropic lets Free, Pro, Max, Team and Enterprise users select Haiku 5.5 on Claude.ai, on the web, iOS and Android.
On the Claude API
New Claude API accounts get a small amount of free credit for testing. After that, you pay the list prices above.
Haiku 5.5 pricing: questions
How much does Claude Haiku 5.5 cost?
$0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, and $0.50 / $2.50 for longer prompts. The Batch API halves both, and cached input costs $0.01 per million tokens.
Is Haiku 5.5 cheaper than Haiku 4.5?
Yes. Per token it is 10× cheaper for prompts up to 100K tokens ($0.10 / $0.50 vs $1 / $5). Haiku 5.5 counts the same text as about 30% more tokens, and Anthropic puts the real saving at 90% for requests up to 100K tokens and 50% above.
Why do prompts over 100K tokens cost more?
Haiku 5.5 has two rate cards. When a request's prompt is over 100,000 tokens, the whole request is billed at $0.50 / $2.50 per million tokens instead of $0.10 / $0.50. The other models in Anthropic's current lineup charge one rate across their 1M-token window.
Do thinking tokens cost extra?
They are billed as output tokens at the normal output rate. Haiku 5.5 thinks adaptively at the default medium effort; choose low effort to spend fewer thinking tokens.
How much does web search cost with Haiku 5.5?
$10 per 1,000 searches on the Claude API, plus the tokens of the search results. Web fetch has no fee beyond the tokens of the fetched content.
Can I use Haiku 5.5 for free?
Yes. On Haiku55 you can chat with it without an account, and signed-in users get free credits every day. Anthropic also offers Haiku 5.5 on the free Claude.ai plan, and new Claude API accounts get a small amount of free test credit.
Keep reading
- Haiku 5.5 vs Opus 5.5Benchmarks, price, speed and when to use each.
- Haiku 5.5 benchmarksEvery published score next to Haiku 4.5, Sonnet 5.5 and GPT-6 Luna.
- Haiku 5.5 vs Haiku 4.5What changed in the new Haiku, including API breaking changes.
- Chat with Haiku 5.5 freeNo sign-up needed. Web search and link reading built in.
Try Claude Haiku 5.5 free, right now
No credit card, nothing to install. Ask your first question in seconds.
Start chatting freeSources
- Claude Docs — Pricing
- Claude Docs — Claude Haiku 5.5 overview
- Anthropic — Introducing Claude Haiku 5.5
- Claude Docs — What's new in Claude Haiku 5.5
- Anthropic — Claude Haiku (availability, customer results)
Haiku55 is an independent service and is not affiliated with Anthropic. Prices and scores are as published by Anthropic and the testers named above, checked on October 8, 2026; see the sources for later changes.