Anthropic model on Crazyrouter
All Anthropic models →

Claude Haiku 5.5 API pricing and integration

Claude Haiku 5.5 API with separate rates for requests up to 100,000 input tokens and longer requests. Both Anthropic Messages and OpenAI-compatible Chat Completions are available.

Use this model via API

This model is available through the paid API. Create an account, add funds and create an API key. The $2 Playground credit is for supported text models only, not API calls.

Sign up for API access
Anthropic nativeCodingVisionTiered pricing
Get API KeyView live pricing

Paste the model ID, live price, base URL and a working request straight into your coding agent.

Anthropic
claude-haiku-5-5
Input / 1M tokens$0.100$0.100
Output / 1M tokens$0.500$0.500
Cached input / 1M tokens$0.010$0.010

Base-tier rates shown; longer requests can cost more. Monthly estimates assume every request stays in the first tier.

Model ID
claude-haiku-5-5
Provider
Anthropic
Endpoint
/v1/chat/completions
Task
chat
Input modalities
Text, image
Output modalities
Text

What is Claude Haiku 5.5?

Use claude-haiku-5-5 for this version; claude-haiku-4-5 is a different model. The important billing boundary is 100,000 input tokens per request. Longer requests use the second tier for input, output and caching, so a monthly total alone does not determine cost. Compare the complete tier table before budgeting document processing, chat or coding workloads. API access requires a funded API balance.

Drop-in OpenAI compatibility

Works with the OpenAI SDKs, LangChain, Cursor, Cline and any tool that accepts a custom base URL.

Streaming and tool use

Streaming, tools and structured output depend on the model and upstream channel. Check the required feature on your channel before production use.

Live usage pricing

The rates come from Crazyrouter's billing table; discounts vary by model.

One account for every model

Compare against Claude, GPT, Gemini, DeepSeek and more without juggling vendor accounts.

Claude Haiku 5.5 API quickstart

Point your OpenAI SDK at https://api.crazyrouter.com/v1 and set the model to claude-haiku-5-5. If your client speaks Anthropic Messages natively, use the second example instead.

cURL
curl https://api.crazyrouter.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $CRAZYROUTER_API_KEY" \
  -d '{
    "model": "claude-haiku-5-5",
    "messages": [{"role": "user", "content": "Explain the tradeoffs of this architecture decision."}],
    "stream": true
  }'
Show OpenAI-compatible migration example
cURL
curl https://api.crazyrouter.com/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: $CRAZYROUTER_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-haiku-5-5",
    "max_tokens": 2048,
    "messages": [{"role": "user", "content": "Explain the tradeoffs of this architecture decision."}],
    "stream": true
  }'

Check the Haiku version and request size

Use claude-haiku-5-5, not the older claude-haiku-4-5 ID. The catalogue lists both Messages and Chat Completions. For document processing, test extraction accuracy and missing-field handling on a small labelled set before increasing batch size; this page does not claim a measured speed or accuracy advantage.

Suggested evaluation request

This request is a reproducible starting point, not a recorded model output or benchmark. API calls require a funded balance. Check errors, output fields and usage on your own account.

cURL
curl https://api.crazyrouter.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $CRAZYROUTER_API_KEY" \
  --data-binary '{
  "model": "claude-haiku-5-5",
  "messages": [
    {
      "role": "user",
      "content": "Extract order_id, currency and amount from: Order CR-108, EUR 49.90; order CR-109, amount missing. Return a JSON array, preserve order IDs, and use null for the missing amount. Do not invent a value."
    }
  ]
}'

Source and scope

Gateway catalogue and billing schedule, reviewed 2026-10-11. This confirms published configuration, not end-to-end feature tests or vendor release claims.

Public gateway catalogueDetailed rates and sources

claude-haiku-5-5: provider price comparison

Input (dark) / output (light), USD per 1M tokens. Listed providers do not identify Crazyrouter upstream routes.

Base-tier rates shown; longer requests can cost more. Monthly estimates assume every request stays in the first tier.

Comparison snapshot from LiteLLM and manually maintained entries; collection date is not independent vendor verification. No cache hits assumed in these estimates. Pricing sources

Reference$0.100 / $0.500Azure$0.100 / $0.500Bedrock$0.100 / $0.500Vertex AI$0.100 / $0.500Crazyrouter$0.100 / $0.500

Estimated workload costs

Base-tier rates shown; longer requests can cost more. Monthly estimates assume every request stays in the first tier.

Chatbot · per month
$23
100k conversations × 800 input / 300 output tokens
Documents · per 1k documents
$1.55
8k input / 1.5k output tokens each
Coding · per developer / month
$0.40
2M input / 400k output tokens

Best use cases

Coding assistants and agent loops with tool calls
Long-document analysis, summarisation and extraction
Customer-facing chat with streaming responses
Evaluation and A/B tests across model vendors
Batch classification, tagging and structured extraction
Technical writing, design docs and code review

Works with developer tools

OpenAI SDKs

Set base_url to https://api.crazyrouter.com/v1 and model to claude-haiku-5-5.

Cursor, Cline, Continue

Add Crazyrouter as a custom OpenAI-compatible provider and pick the model from the list.

LangChain / LlamaIndex

Use the OpenAI chat model class with the Crazyrouter base URL.

Claude Code / Codex CLI

CLI compatibility requires the native protocol and a supported model/channel. Follow the integration guide and verify compatibility before use.

Claude Haiku 5.5 FAQ

What base URL do I use for Claude Haiku 5.5?

https://api.crazyrouter.com/v1. Chat completions go to /v1/chat/completions; native Anthropic Messages requests use their usual path.

How is usage billed?

Per token, at the input/output rates shown on this page. Your balance is debited per request; there is no subscription. Base-tier rates shown; longer requests can cost more. Monthly estimates assume every request stays in the first tier.

Does it support streaming and function calling?

Support varies by model and upstream channel. The request example is a starting point; verify streaming, tools and structured output separately before production use.

Can I switch to another model later?

Change the model string. The same key works for every model in the Crazyrouter catalogue.

Where do I see the current price?

The pricing page lists live rates for every model, including group discounts if you are logged in.