Skip to content
AZRouter.dev

Anthropic · Text model

Anthropic: Claude Haiku 5.5

anthropic/claude-haiku-5.51M context

Claude Haiku 5.5 is Anthropic's small, fast model for high-volume, cost-sensitive work such as summarization, subagents, and browser use.

Modalities
Price vs OpenRouter
$0.11/M inOpenRouter $0.10+10%$0.55/M outOpenRouter $0.50+10%Tiered above 100K
Context
1M
Max output
128K
Released
Oct 2026

Model specifications

Provider
Anthropic
Model slug
anthropic/claude-haiku-5.5
Input
Text, Image, File
Output
Text
Context length
1,000,000 tokens
Max output
128,000 tokens
Long context
Tiered above 100Kthe whole request is repriced; see Pricing
Released
Oct 7, 2026
Price reference
OpenRouter public price, checked

Pricing

Every unit this model bills by, OpenRouter's price next to AZRouter's. Long prompts move the whole request to the long-context rates.

  • Input /M tokprompt up to 100K tokens

    +10%
    OpenRouter
    $0.10
    AZRouter
    $0.11

  • Input /M tokprompt over 100K tokens

    +10%
    OpenRouter
    $0.50
    AZRouter
    $0.55

  • Output /M tokprompt up to 100K tokens

    +10%
    OpenRouter
    $0.50
    AZRouter
    $0.55

  • Output /M tokprompt over 100K tokens

    +10%
    OpenRouter
    $2.50
    AZRouter
    $2.75

  • Cache read /M tokprompt up to 100K tokens

    +10%
    OpenRouter
    $0.01
    AZRouter
    $0.011

  • Cache read /M tokprompt over 100K tokens

    +10%
    OpenRouter
    $0.05
    AZRouter
    $0.055

  • Cache write /M tokprompt up to 100K tokens

    +10%
    OpenRouter
    $0.125
    AZRouter
    $0.1375

  • Cache write /M tokprompt over 100K tokens

    +10%
    OpenRouter
    $0.625
    AZRouter
    $0.6875

A request whose prompt is over 100,000 tokens is billed entirely at the long-context rates, output and cache included. The threshold and both rates are the provider's published pricing.

The prompt cache TTL is 5 minutes.

A request that asks for a one-hour cache is served with the 5-minute cache and billed as a 5-minute cache write, at the cache write price above. Only cache_control markers on content blocks, in system or a message's content, cache; a marker at the top level of the request or on a tool is dropped. Cached prompt tokens count toward the long-context threshold. How Claude caching is billed

Reference: OpenRouter's public price, checked . You are billed the AZRouter price from prepaid credits.

Each request's total is rounded to the nearest $0.000002 (halves up), so it can differ from the per-unit math by up to $0.000001. How rounding works

API

OpenAI-compatible. Base URL https://api.azrouter.dev/v1, key in $AZROUTER_API_KEY.

curl "https://api.azrouter.dev/v1/chat/completions" \
  -H "Authorization: Bearer $AZROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-haiku-5.5",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Requests may be used for model training by AZRouter, upstream providers and model owners. Privacy Policy

Try this model