Skip to content
Anthropic logo

Claude Haiku 5.5

Text Generation • Anthropic

View as MarkdownAgent setup

Model ID

Use with the Workers AI binding, the AI Gateway REST API, or the Unified API.

Calling the provider directly? Use the provider's model name instead, including with provider-native endpoints or the provider's own SDK. For example, Anthropic's name for anthropic/claude-opus-4.8 is claude-opus-4-8.

  • Third-party

Claude Haiku 5.5 is a fast, cost-efficient model for high-volume, latency-sensitive tasks such as classification, extraction, routing, and subagent work.

Model Info
Context Window ↗1,000,000 tokens
Terms and Licenselink ↗
More informationlink ↗
Request formatsAnthropic Messages
Pricing
  • Short-context input (per 1M)$0.10
  • Short-context cached input (per 1M)$0.01
  • Short-context cache write (per 1M)$0.125
  • Short-context output (per 1M)$0.50
  • Long-context input (per 1M)$0.50
  • Long-context cached input (per 1M)$0.05
  • Long-context cache write (per 1M)$0.625
  • Long-context output (per 1M)$2.50

Usage

const response = await env.AI.run(
  'anthropic/claude-haiku-5.5',
  {
    max_tokens: 256,
    messages: [{ content: 'Classify this support ticket as billing, technical, or account: My invoice shows an unexpected charge.', role: 'user' }],
  },
)
console.log(response)
curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/v1/messages \
  --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "anthropic/claude-haiku-5.5",
  "max_tokens": 256,
  "messages": [
    {
      "content": "Classify this support ticket as billing, technical, or account: My invoice shows an unexpected charge.",
      "role": "user"
    }
  ]
}'
**Billing**

The ticket is about an invoice showing an unexpected charge, which falls under billing.
{
  "id": "msg_011CfoNwxiMXWCayWiWZRotX",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "**Billing**\n\nThe ticket is about an invoice showing an unexpected charge, which falls under billing."
    }
  ],
  "model": "claude-haiku-5-5",
  "stop_reason": "end_turn",
  "usage": {
    "input_tokens": 40,
    "output_tokens": 35
  },
  "container": null,
  "stop_sequence": null,
  "stop_details": null,
  "diagnostics": null
}

Examples

Structured Extraction — Use adaptive thinking for structured extraction.
const response = await env.AI.run(
  'anthropic/claude-haiku-5.5',
  {
    max_tokens: 1024,
    messages: [{ content: 'Extract the order ID and requested action from: Please cancel order #A-1042 before it ships.', role: 'user' }],
    output_config: { effort: 'medium' },
    thinking: { type: 'adaptive' },
  },
)
console.log(response)
curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/v1/messages \
  --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "anthropic/claude-haiku-5.5",
  "max_tokens": 1024,
  "messages": [
    {
      "content": "Extract the order ID and requested action from: Please cancel order #A-1042 before it ships.",
      "role": "user"
    }
  ],
  "output_config": {"effort": "medium"},
  "thinking": {"type": "adaptive"}
}'
**Order ID:** A-1042
**Requested action:** Cancel the order before it ships
{
  "id": "msg_011CfoNx2fFMc8rnUuBBgE3A",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "**Order ID:** A-1042\n**Requested action:** Cancel the order before it ships"
    }
  ],
  "model": "claude-haiku-5-5",
  "stop_reason": "end_turn",
  "usage": {
    "input_tokens": 38,
    "output_tokens": 31
  },
  "container": null,
  "stop_sequence": null,
  "stop_details": null,
  "diagnostics": null
}
Streaming Response — Stream a low-latency response incrementally.
const response = await env.AI.run(
  'anthropic/claude-haiku-5.5',
  {
    max_tokens: 512,
    messages: [{ content: 'Return a concise three-step deployment checklist.', role: 'user' }],
    stream: true,
  },
)
console.log(response)
curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/v1/messages \
  --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "anthropic/claude-haiku-5.5",
  "max_tokens": 512,
  "messages": [{"content": "Return a concise three-step deployment checklist.", "role": "user"}],
  "stream": true
}'
**Three-Step Deployment Checklist**

1. **Pre-deployment**
   - Confirm all tests pass and code is reviewed and approved.
   - Verify configuration, environment variables, and secrets are set for the target environment.
   - Back up the current production state and prepare a rollback plan.

2. **Deploy**
   - Release to production using the approved pipeline or process, avoiding manual changes where possible.
   - Run database migrations or other required schema changes, if any.
   - Monitor the rollout in stages if supported (e.g., canary or gradual release).

3. **Post-deployment**
   - Run smoke tests on critical paths.
   - Check logs, error rates, and performance metrics against baseline.
   - Notify stakeholders of completion and keep the rollback plan ready for a defined monitoring window.
[
  {
    "type": "message_start",
    "message": {
      "model": "claude-haiku-5-5",
      "id": "msg_011CfoNx62eTigAN2u1rkUKg",
      "type": "message",
      "role": "assistant",
      "content": [],
      "container": null,
      "stop_reason": null,
      "stop_sequence": null,
      "stop_details": null,
      "usage": {
        "input_tokens": 23,
        "cache_creation_input_tokens": 0,
        "cache_read_input_tokens": 0,
        "cache_creation": {
          "ephemeral_5m_input_tokens": 0,
          "ephemeral_1h_input_tokens": 0
        },
        "output_tokens": 4,
        "service_tier": "standard",
        "inference_geo": "global"
      },
      "diagnostics": null
    }
  },
  {
    "type": "content_block_start",
    "index": 0,
    "content_block": {
      "type": "thinking",
      "thinking": "",
      "signature": ""
    }
  },
  "... 75 more chunks omitted ...",
  {
    "type": "message_stop"
  }
]

Parameters

max_tokens
numberrequiredexclusiveMinimum: 0
stream
boolean
id
string
type
stringconst: message
role
stringconst: assistant
model
string
stop_reason
string

API Schemas (Raw)

Input
Output

Was this helpful?