Claude Haiku 5.5 for classification and extraction
Use Claude Haiku 5.5 for classification and document extraction. $0.35 / 1M input, $1.75 / 1M output.
- Adaptive thinking
- 1M-token context
- Text and image input
- Tool use
- Streaming
Try Claude Haiku 5.5
Test Claude Haiku 5.5 with your own tasks. Inspect answers and token usage.
claude-haiku-5-5$0.35 in · $1.75 out per 1M tokens
Disabled thinking supports low, medium or high effort. Check beta requirements in the reference.
Claude Haiku 5.5 output
Unedited result at default effort.
Classify this support request as billing, technical or account: I was charged twice for my subscription. Return only the label.
Billing
October 9, 2026: 41 input, 5 output tokens. Results vary.
Claude Haiku 5.5 pricing
Live Claude Haiku 5.5 token rates, refreshed automatically.
Illustrative sizes; charges follow reported usage.
| Category | Price · 1M tokens |
|---|---|
| Input | $0.35 |
| Output | $1.75 |
| Cache read | $0.035 |
| Cache write, 5 minutes | $0.4375 |
What is Claude Haiku 5.5?
Claude Haiku 5.5 is Anthropic's model for classification, extraction and summaries. Send documents or messages; get text or tool requests.
Claude Haiku 5.5 defaults to medium-effort adaptive thinking. Compare settings on your tasks.
Released October 7, 2026 · Knowledge cutoff: June 2026
Connect your app or coding agent
Connect Claude Haiku 5.5 to your app or agent.
Call it from your backend
Call Claude Haiku 5.5 with the Anthropic SDK and your SeedRouter key.
- 1Create an API key
- 2Set https://api.seedrouter.ai as base URL
- 3Choose claude-haiku-5-5
Hand it to your coding agent
Ask your agent to prepare a Claude Haiku 5.5 request and cost estimate for approval.
- 1Copy the prompt
- 2Paste it into your agent
- 3Approve the request
Work with messages and documents
Give Claude Haiku 5.5 a focused task with a checkable result.
Classify incoming text
Sort messages into your own labels with Claude Haiku 5.5.
Summarize longer material
Ask Claude Haiku 5.5 for a summary, then check it against the source.
Extract consistent fields
Extract dates and quantities with Claude Haiku 5.5. Define how missing values should appear.
Delegate a focused task
Give Claude Haiku 5.5 a bounded task, inspect tool arguments and return results.
Adjust thinking to the job
Compare Claude Haiku 5.5 effort settings on the same examples.
Reuse a stable prompt
Cache repeated Claude Haiku 5.5 instructions for five minutes or one hour.
Build around a defined workflow
Feedback labeling
Label feedback with Claude Haiku 5.5. Keep a reviewed test set for prompt changes.
Support drafts
Draft Claude Haiku 5.5 replies from approved articles. Review before sending.
Document extraction
Extract JSON with Claude Haiku 5.5; verify the values against the document.
Start with Claude Haiku 5.5
- 01
Create a key
Create a key and add credit for Claude Haiku 5.5.
- 02
Set the base URL
Anthropic: https://api.seedrouter.ai. OpenAI: append /v1.
- 03
Choose your model
Set claude-haiku-5-5 and max_tokens for your Claude Haiku 5.5 request.
- 04
Review usage
Review Claude Haiku 5.5 answers and tokens in usage records.
Claude Haiku 5.5 at a glance
Tested capabilities and limits are in the API reference.
- Anthropic Messages API
- Native Claude request format
- OpenAI Chat Completions
- Basic and reasoning requests verified
- OpenAI Responses
- Basic and reasoning requests verified
- Image and PDF input
- Base64 image and PDF examples verified
- Prompt caching
- 5-minute and 1-hour
- Forced tool use
- any and named tool requests verified
- Failed requests
- Not charged
Claude Haiku 5.5 batch processing and fast mode are not offered here.
Limits and compatibility to check
Context and output
Claude Haiku 5.5 lists 1M context and 128K output; these ceilings remain untested here.
Thinking restrictions
Claude Haiku 5.5 officially supports disabled thinking; some tests still returned thinking blocks.
Sampling restrictions
Omit sampling settings for Claude Haiku 5.5. Use effort to adjust reasoning.
Zero-output requests
Claude Haiku 5.5 passed zero-output prewarming. Streaming, structured output and forced tools cannot accompany it.
Context management
Claude Haiku 5.5 returned a signed compaction block. Follow beta and history requirements in the reference.
Instructions and history
Claude Haiku 5.5 system-instruction and per-message effort tests found inconsistencies.
Why use SeedRouter?
One key
Use one SeedRouter key for Claude Haiku 5.5 and other available models.
Pay as you go
Spend prepaid credit on Claude Haiku 5.5 without a subscription.
Failures are free
Claude Haiku 5.5 requests that return errors are not charged.
Live prices
Claude Haiku 5.5 prices come from current billing rates.
Prompt caching
Claude Haiku 5.5 passed both cache-usage checks. Official minimum: 512 tokens.
Images and PDFs
Image and PDF examples passed with Claude Haiku 5.5; full media boundaries remain unverified.
Streaming
Stream Claude Haiku 5.5 output; handle thinking, tools and errors.
Related models
Compare Claude Haiku 5.5 with other available models.
Call Claude Fable 5 with the Anthropic or OpenAI SDK. Deep reasoning, 1M-token context, forced tool use, live per-token pricing and a playground to test it.
View pricingCall Claude Fable 5.1 with the Anthropic or OpenAI SDK. Deep reasoning, 1M-token context, cheap cached reads, live pricing and a playground to try it today.
View pricingCall Claude Opus 5.5 with the Anthropic or OpenAI SDK. 1M-token context, adaptive thinking, live per-token pricing and a playground you can try right now.
View pricingTry Claude Sonnet 5.5 for code reviews and document questions. See live token prices, test a prompt, and check verified capabilities before connecting your app.
View pricingCall DeepSeek V4.1 Flash with the OpenAI or Anthropic SDK. 1M-token context, thinking on or off, image input, live peak and off-peak pricing and a playground.
View pricingGenerate video with Seedance 2.0: text, first and last frames, and image, video or audio references. 480p to 4K, 4–15 seconds. Pay per second, no subscription.
View pricingClaude Haiku 5.5 guides
Understand token costs and compare Haiku with Sonnet.
Understand Claude Haiku 5.5 token prices, the 100K input threshold, cache writes and reads, and how to estimate a completed request.
Compare Claude Haiku 5.5 and Sonnet 5.5 by task, thinking defaults and tool behavior, then build a small evaluation around your actual workload.
Claude Haiku 5.5 FAQ
What is Haiku 5.5?+
Claude Haiku 5.5 is Anthropic's model for classification, extraction and summaries. It accepts text, images and documents, returning text or tool requests.
Should I use Haiku 5.5 or Sonnet 5.5?+
Compare Claude Haiku 5.5 with Sonnet on your tasks. Check answer quality and total usage.
How much does Haiku 5.5 cost?+
Claude Haiku 5.5 costs $0.35 / 1M input and $1.75 / 1M output on SeedRouter. Cache rates appear above; charges follow reported tokens.
Can I disable thinking?+
Claude Haiku 5.5 officially supports disabled thinking at low, medium or high effort. Tests still returned thinking in some cases; check compatibility notes.
When was Haiku 5.5 released?+
Claude Haiku 5.5 launched October 7, 2026, with a June 2026 knowledge cutoff. Supply current sources when needed.
Which model ID do I use?+
Use claude-haiku-5-5 for Claude Haiku 5.5. Messages requires model, max_tokens and messages. Keep keys server-side.
Can I use the OpenAI SDK?+
Claude Haiku 5.5 Chat Completions and Responses passed basic and reasoning tests. Use Messages for native Claude parameters.
Are failed requests charged?+
Claude Haiku 5.5 requests that return errors are not charged. Successful responses stopped at the output limit are charged for reported usage.
Put Claude Haiku 5.5 to work
Try Claude Haiku 5.5 on a task before connecting your app.