> ## Documentation Index > Fetch the complete documentation index at: https://docs.anyfast.ai/llms.txt > Use this file to discover all available pages before exploring further. # claude-haiku-5-5 > Create Claude Haiku 5.5 Messages through AnyFast. The tested AnyFast response reported `model: "claude-haiku-5.5"` for a `claude-haiku-5-5` request and added `request_id` plus `billing`. When disabling thinking, omit `output_config.effort`; sending explicit low effort caused a `400` conflict. ## OpenAPI ````yaml api-reference/model-api/anthropic/openapi/claude-haiku-5-5/openapi.yaml POST /v1/messages openapi: 3.1.0 info: title: Claude Haiku 5.5 description: >- Use Anthropic Claude Haiku 5.5 through AnyFast. The model has a 1M-token context window, supports up to 128K output tokens, and uses adaptive thinking by default. version: 1.0.0 servers: - url: https://www.anyfast.ai security: - bearerAuth: [] paths: /v1/messages: post: summary: Create a Message description: > Send text, images, prior assistant turns, and tool results. Claude Haiku 5.5 can return text, thinking, and tool-use content blocks. Adaptive thinking is on by default. Use `output_config.effort` to control depth. `thinking.type=disabled` is available through high effort. Forced tool choice is accepted but starts with a tool call without a thinking block. operationId: createMessageClaudeHaiku55 requestBody: required: true content: application/json: schema: type: object required: - model - max_tokens - messages properties: model: type: string enum: - claude-haiku-5-5 description: Model ID. example: claude-haiku-5-5 messages: type: array minItems: 1 description: >- Conversation turns. Pass assistant content blocks, including thinking blocks, back unchanged in multi-turn conversations. Do not prefill the final assistant response. items: type: object required: - role - content properties: role: type: string enum: - user - assistant description: Message author. content: oneOf: - type: string - type: array items: type: object properties: type: type: string enum: - text - image - tool_result - thinking - tool_use text: type: string source: type: object properties: type: type: string enum: - base64 - url media_type: type: string data: type: string url: type: string format: uri tool_use_id: type: string content: oneOf: - type: string - type: array items: type: object is_error: type: boolean thinking: type: string signature: type: string id: type: string name: type: string input: type: object additionalProperties: true description: Text or an array of content blocks. example: - role: user content: >- Identify the highest-risk assumptions in this migration plan. max_tokens: type: integer minimum: 1 maximum: 128000 description: >- Maximum output tokens across thinking and visible response text. example: 4096 system: oneOf: - type: string - type: array items: type: object properties: type: type: string enum: - text text: type: string description: >- System prompt. Keep it unchanged when replaying Haiku 5.5 thinking blocks. metadata: type: object properties: user_id: type: string description: >- External identifier for the user associated with the request. description: Request metadata. stop_sequences: type: array items: type: string description: Custom text sequences that stop generation. stream: type: boolean default: false description: Stream message events over SSE. tools: type: array description: >- Tool definitions available to the model. Keep this array unchanged when replaying Haiku 5.5 thinking blocks. items: type: object required: - name - input_schema properties: name: type: string description: Tool name. description: type: string description: What the tool does and when to use it. input_schema: type: object description: JSON Schema for the tool input. additionalProperties: true tool_choice: type: object required: - type properties: type: type: string enum: - auto - none - any - tool default: auto description: >- `auto` lets Claude decide whether to call a tool; `none` prevents tool calls. Forced `any` and named `tool` choices start with a tool call and omit a preceding thinking block. name: type: string description: Tool name when type is `tool`. description: Controls whether the model may use tools. output_config: type: object properties: effort: type: string enum: - low - medium - high - xhigh - max default: medium description: >- Controls adaptive-thinking depth, latency, and token use. description: Output controls. Claude Haiku 5.5 defaults to medium effort. thinking: type: object properties: type: type: string enum: - adaptive - disabled default: adaptive description: >- Adaptive is the default; disabled is supported at high effort or below. description: >- Omit this field for adaptive thinking. For AnyFast, omit `output_config.effort` with `disabled`; an explicit low effort returned a conflict error. Manual `enabled` with `budget_tokens` returns 400. responses: '200': description: >- Message created successfully. A safeguard refusal is also returned with HTTP 200 and `stop_reason` set to `refusal`. content: application/json: schema: $ref: '#/components/schemas/Message' '400': description: >- Invalid request, including unsupported thinking modes, assistant prefilling, or invalidated thinking blocks. content: application/json: schema: $ref: '#/components/schemas/Error' '401': description: Unauthorized content: application/json: schema: $ref: '#/components/schemas/Error' '429': description: Rate limit exceeded content: application/json: schema: $ref: '#/components/schemas/Error' x-codeSamples: - lang: bash label: cURL source: | curl https://www.anyfast.ai/v1/messages \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "claude-haiku-5-5", "max_tokens": 4096, "output_config": { "effort": "medium" }, "messages": [ { "role": "user", "content": "Identify the highest-risk assumptions in this migration plan." } ] }' components: schemas: Message: type: object required: - id - type - role - content - model - stop_reason - usage properties: id: type: string example: msg_01XFDUDYJgAACzvnptvVoYEL type: type: string enum: - message role: type: string enum: - assistant content: type: array items: type: object properties: type: type: string enum: - text - thinking - tool_use text: type: string thinking: type: string signature: type: string id: type: string name: type: string input: type: object additionalProperties: true example: - type: text text: >- The highest-risk assumption is that every client can tolerate the planned cutover window. model: type: string example: claude-haiku-5.5 description: >- The observed AnyFast response uses a dot-form ID; requests still use the hyphen-form public model ID. container: type: object nullable: true additionalProperties: true description: Null in the observed AnyFast text response. stop_reason: type: string enum: - end_turn - max_tokens - stop_sequence - tool_use - refusal - model_context_window_exceeded - null stop_sequence: type: string nullable: true stop_details: type: object nullable: true additionalProperties: true description: Additional details for a refusal or other stop condition. request_id: type: string description: Request ID added by AnyFast in the response body. billing: type: object description: AnyFast billing metadata; the billed USD amount is a string. properties: status: type: string billed_cost_usd: type: string currency: type: string zero_data_retention: type: boolean effective_discount_percent: type: number usage: type: object properties: service_tier: type: string example: standard input_tokens: type: integer output_tokens: type: integer cache_creation_input_tokens: type: integer cache_read_input_tokens: type: integer cache_creation: type: object properties: ephemeral_1h_input_tokens: type: integer ephemeral_5m_input_tokens: type: integer output_tokens_details: type: object properties: thinking_tokens: type: integer Error: type: object properties: type: type: string example: error error: type: object properties: type: type: string example: invalid_request_error message: type: string example: Invalid request parameters securitySchemes: bearerAuth: type: http scheme: bearer ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.