ClaudeAnthropic Repeatable classification and extraction

Claude Haiku 4.5

Sort a request into a fixed intent taxonomy or read one visible form field.

Anthropic lists Haiku 4.5 as an active legacy model and documents manual extended thinking for its direct API. Use this TokenHot route to evaluate the workflows below against your own acceptance criteria.

  • Text and image input; text output
  • 200K-token catalog context

The published TokenHot Messages example sends text. The chat Playground accepts one new image attachment per message. Confirm media and optional settings for the API endpoint you use.

AI Chat

Send a prompt and see the response here.

claude-haiku-4-5
Current conversation
How can I help you?

Start a conversation with this model and ask follow-up questions.

Uses your account balance at the model’s current rates.

USD /1M tokens

Pricing

GroupInputOutputCache ReadCache Write
default$1$5$0.11$1.375
claude-0.34$0.34$1.7$0.0374$0.4675
Claud0.7$0.7$3.5$0.077$0.9625
default
Input$1Output$5Cache Read$0.11Cache Write$1.375
claude-0.34
Input$0.34Output$1.7Cache Read$0.0374Cache Write$0.4675
Claud0.7
Input$0.7Output$3.5Cache Read$0.077Cache Write$0.9625

Public list prices are shown in USD. Final charges may vary by account group and usage tier.

Overview

What is Claude Haiku 4.5?

Anthropic lists Haiku 4.5 as an active legacy model and documents manual extended thinking for its direct API.

Supply the material that matters, label each source and ask for an answer with a fixed format. Check the result against the originals before relying on it.

intent sort

Sort a request into a fixed intent taxonomy

An allowed label and an escalation flag for ambiguous requests.

field review

Read one visible form field

A single visible value or an explicit unknown result.

ticket summary

Summarize a resolved ticket in a fixed format

A compact issue, action and outcome record.

From brief to deliverable

Put Claude Haiku 4.5 to work

Start with a concrete task and decide what a useful result looks like.

Repeatable classification and extraction

Sort a request into a fixed intent taxonomy

Use one short request, allowed labels and policy examples to draft a result a teammate can verify.

Tools you need
Labeled source material and a reviewer with access to the originals.
Keep in mind
Keep final routing and validation in your application.
View a task brief

Sort a request into a fixed intent taxonomy. Use only the supplied evidence. Return an allowed label and an escalation flag for ambiguous requests. Label any missing information.

Repeatable classification and extraction

Read one visible form field

Use one clear form screenshot and the field name to draft a result a teammate can verify.

Tools you need
Labeled source material and a reviewer with access to the originals.
Keep in mind
Use the original record to verify a value before storage.
View a task brief

Read one visible form field. Use only the supplied evidence. Return a single visible value or an explicit unknown result. Label any missing information.

Repeatable classification and extraction

Summarize a resolved ticket in a fixed format

Use the ticket ID and closing notes to draft a result a teammate can verify.

Tools you need
Labeled source material and a reviewer with access to the originals.
Keep in mind
Keep missing information open and do not infer that the issue was resolved.
View a task brief

Summarize a resolved ticket in a fixed format. Use only the supplied evidence. Return a compact issue, action and outcome record. Label any missing information.

Capabilities in context

Key features

FeatureWhat you getWhy it matters
TokenHot model IDclaude-haiku-4-5Use the exact value from the model page.
Catalog modalitiesText and image input; text outputCheck media payloads against your API endpoint.
Catalog context200K tokensLeave room in the request for the written response.
TokenHot API exampleText through /v1/messagesStart with the published request shape.
PlaygroundOne new image attachment per chat messageUse a single image for a visual task.

TokenHot's catalog and example describe this route. Anthropic's documentation describes the Claude API. Images were made with ImageGen to illustrate the workflows. TokenHot model details · Anthropic model information · Anthropic context windows

Audience & fit

Where Claude Haiku 4.5 fits

Sort a request into a fixed intent taxonomy

An allowed label and an escalation flag for ambiguous requests.

Read one visible form field

A single visible value or an explicit unknown result.

Summarize a resolved ticket in a fixed format

A compact issue, action and outcome record.

Cost context

Plan the cost of a useful result

Check the current TokenHot rates

Use the pricing table for this route and the group available to your account.

Measure representative input and output

Use the request sizes and answer formats the application actually sends.

Compare accepted results

Include retries and review effort when comparing routes.

Use the current pricing table for available rates and account groups. Model IDs and prices come from TokenHot’s catalog; provider pricing and subscriptions are separate.

API

Code examples

curl --location --request POST 'https://api.tokenhot.ai/v1/messages' \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data-raw '{
    "model": "claude-haiku-4-5",
    "messages": [
        {
            "role": "user",
            "content": "What can you do?"
        }
    ],
    "max_tokens": 100
}'

Blog

Related Articles

FAQ

Frequently asked questions

What is Claude Haiku 4.5 best suited for?

Claude Haiku 4.5 is suited to the input, output, and capability types listed on this page. Test production workloads before choosing it for a critical system.

How is Claude Haiku 4.5 priced?

Pricing is shown in the pricing section above. Actual cost depends on usage volume, account group, and the request payload.

How can I call Claude Haiku 4.5?

Use the model ID shown above with one of the supported API protocols. Code examples are provided when usage data is available.

Which model ID should I use?

Use claude-haiku-4-5 in the model field.

What does the TokenHot API example show?

It sends text to /v1/messages. Confirm other fields and media payloads before automating them.

Can I attach a screenshot in the Playground?

Yes. The chat Playground accepts one new image per message and returns text.

What context window does TokenHot list?

The catalog lists 200K tokens for this route.

When should I compare with Claude Haiku 5.5?

Run both route IDs with the same inputs and acceptance criteria. Compare quality, cost and reviewer effort before switching.

Get started

Build with Claude Haiku 4.5

Use in Console