GPT Image 2.5 is live — OpenAI's newest image model, targeted edits that leave the rest of the frame alone

Claude Haiku 5.5 — Writing, Extraction & Tool Use on reAPI

Use Claude Haiku 5.5 to classify messages, extract structured records, and draft replies from your instructions. Test your own workflow on reAPI.

Highlights
1M context · adaptive thinking · medium default effort
Best for
Classification, structured extraction, everyday writing
Input
Text and images
Output
Text, structured JSON, tool calls
Claude Haiku 5.5

Ask Claude Haiku 5.5

Start with the prompt in the composer or choose an example.

Enter to send · Shift+Enter for a new line

Pricing

Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.

30% off official pricing· Input / output tokens

First test cost
$0.00212.1 credits

One 1K-token prompt with a 1K-token answer at the cheapest tier.

Testing budget guide
Add credits
$10
≈ 4761 tests
$50
≈ 23809 tests
$100
≈ 47619 tests
Input
$0.3500
Price · 1M tokens
Output
$1.7500
Price · 1M tokens
Cache read
$0.0350
Price · 1M tokens
Cache write
$0.4375
Price · 1M tokens

What is Claude Haiku 5.5?

Claude Haiku 5.5 is an Anthropic language model for everyday tasks. It takes text and images, supports up to one million tokens of context, and returns text or structured results. Use Claude Haiku 5.5 to turn requests into decisions your team can review and build a repeatable workflow.

What you can build with this model

Real-world workflows and production use cases you can build and ship with this model.

Claude Haiku 5.5 product poster showing messages sorted into work queues

Claude Haiku 5.5 for routing incoming requests

Give Claude Haiku 5.5 the categories your team actually uses, together with a few examples of ambiguous requests. Ask for a category and the sentence that supports it, so someone can check the decision without reading the whole conversation. Keep an explicit review path for messages that lack enough information.

Try Claude Haiku 5.5
Claude Haiku 5.5 product poster showing source documents beside structured fields

Claude Haiku 5.5 for structured extraction

Ask Claude Haiku 5.5 to turn a supplier note into fields your application can use. Claude Haiku 5.5 supports structured JSON output, including explicit missing values. Check extracted names and dates against the original before saving them.

Claude Haiku 5.5 product poster showing reference cards and a finished draft

Claude Haiku 5.5 for practical first drafts

Give Claude Haiku 5.5 the policy, customer question, and preferred tone. Claude Haiku 5.5 can draft a reply explaining the next step or condense a conversation into a handoff note. Set a clear length and leave unresolved questions visible.

Why reAPI for Claude Haiku 5.5

Evaluate Claude Haiku 5.5 on real tasks

Start with a few requests whose correct outcomes you already know. Run Claude Haiku 5.5 in the playground, review the results, and carry those prompts into your application once they meet your requirements.

Keep Claude Haiku 5.5 costs visible

Estimate your Claude Haiku 5.5 workload using the current rates above. Keep responses concise for repeated classification and extraction, then check actual usage as you expand.

Connect Claude Haiku 5.5 to your tools

Give Claude Haiku 5.5 a tool definition with the fields your application expects. Claude Haiku 5.5 supports forced tool selection when a workflow needs that particular tool to answer.

Claude Haiku 5.5 vs Claude Sonnet 5.5

The two models share a large context window but differ in their default effort and tool controls. Use these documented differences to choose an initial candidate, then compare the quality and cost of your own tasks.

Documented difference
Claude Haiku 5.5
Claude Sonnet 5.5
Default effort
Medium
High
Skip initial thinking
Disabled at low, medium, or high effort
Between-tools mode at low, medium, or high effort
Force a named tool
Supported
Use automatic selection
Context window
1 million tokens
1 million tokens
Standard output ceiling
128,000 tokens, including thinking
128,000 tokens, including thinking
Official release
October 7, 2026
September 28, 2026

Based on Anthropic model and migration documentation checked October 9, 2026. reAPI prices are shown separately above. These specifications are not a benchmark ranking.

Start using Claude Haiku 5.5

  1. step 01

    Get a key for Claude Haiku 5.5

    Sign in, open API keys, and create a credential for your application. Keep it on your server and add funds before making paid requests.

    Open
  2. step 02

    Send a Claude Haiku 5.5 message

    Use claude-haiku-5-5 with the Messages endpoint at https://reapi.ai/api/v1/messages. Follow the documentation example, then replace the sample question with one task you can check.

    Open
  3. step 03

    Review your Claude Haiku 5.5 result

    Compare the answer with your expected result, including any fields that should remain unknown. Save the prompt and output requirements so future requests use the same evaluation criteria.

    Open
docs/api/claude-haiku-5-5

API reference

Drop-in code and the full parameter table.

curl --no-buffer --request POST \
  --url https://reapi.ai/api/v1/messages \
  --header "x-api-key: YOUR_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "claude-haiku-5-5",
  "messages": [
    {
      "role": "user",
      "content": "Explain this design clearly."
    }
  ],
  "max_tokens": 4096,
  "output_config": {
    "effort": "medium"
  },
  "stream": true
}'

Check the response

Read stop_reason and returned content, including incomplete answers or refusals. HTTP 200 alone does not verify every parameter.

Authentication and balance

Use an active reAPI key and a funded account. The docs explain request authentication and token billing.

Continue tool conversations

Return the tool result with its matching ID and preserve the complete assistant content from the previous turn.

Handle upstream errors

Inspect the returned error and request ID. Decide whether a retry is appropriate for that error and your application.

Frequently asked questions

Common questions about this model.

Claude Haiku 5.5 reads text and images and returns text. You can use it for classification, extraction, summaries, and drafts, or request structured output for an application. Give it the source material and the result you need, then check the response against that material.

start building

Ready to ship?

Try it in the playground or grab an API key to integrate now.