> ## Documentation Index > Fetch the complete documentation index at: https://docs.synapsai.cloud/llms.txt > Use this file to discover all available pages before exploring further. # Create a model response > OpenAI-compatible Responses API backed by vLLM deployments only. Provide text or structured input to generate a response with optional tool calling, reasoning, and streaming. **Availability:** Only supported on vLLM-backed model deployments. **Stateless operation:** Responses are not stored server-side. The `store` parameter is always forced to `false`. Use `previous_input_messages` to carry conversation history across turns instead of `previous_response_id`. **Unsupported parameters:** Requests that set `previous_response_id` or `background: true` are rejected at validation with a clear error. **Batching:** This endpoint does not support batch requests. ## OpenAPI ````yaml /api-reference/openapi.json post /v1/responses openapi: 3.1.0 info: title: SynapsAI Inference API version: 1.0.0 servers: - url: https://api.synapsai.cloud description: Production server security: [] paths: /v1/responses: post: summary: Create a model response description: >- OpenAI-compatible Responses API backed by vLLM deployments only. Provide text or structured input to generate a response with optional tool calling, reasoning, and streaming. **Availability:** Only supported on vLLM-backed model deployments. **Stateless operation:** Responses are not stored server-side. The `store` parameter is always forced to `false`. Use `previous_input_messages` to carry conversation history across turns instead of `previous_response_id`. **Unsupported parameters:** Requests that set `previous_response_id` or `background: true` are rejected at validation with a clear error. **Batching:** This endpoint does not support batch requests. requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/ResponsesRequest' responses: '200': description: Model response content: application/json: schema: $ref: '#/components/schemas/ResponsesResponse' '400': description: >- Invalid request (for example, unsupported `previous_response_id` or `background` mode) components: schemas: ResponsesRequest: allOf: - type: object description: >- Request body for the OpenAI-compatible Responses API. The primary input field is `input`. Only available on vLLM-backed deployments. Responses are not persisted; `store` is always forced to `false`. properties: model: type: string description: Model ID to use for this request. Required. input: oneOf: - type: string description: A text input to the model. - type: array items: type: object additionalProperties: true description: >- A list of input items (for example, role/content message objects). description: Text or structured input to the model. Required. instructions: type: string nullable: true description: >- System or developer instructions inserted into the model context, separate from `input`. max_output_tokens: type: integer nullable: true description: Maximum number of tokens the model may generate in the response. max_tool_calls: type: integer nullable: true description: >- Maximum number of tool calls the model may make during generation. metadata: type: object additionalProperties: true nullable: true description: Optional key-value metadata attached to the request. logit_bias: type: object additionalProperties: type: number nullable: true description: Mapping from token ID to a bias value added to the token logits. parallel_tool_calls: type: boolean default: true description: Whether the model may call multiple tools in parallel. prompt: type: object additionalProperties: true nullable: true description: Optional prompt template configuration. reasoning: type: object additionalProperties: true nullable: true description: Optional reasoning configuration for supported models. service_tier: type: string enum: - auto - default - flex - scale - priority default: auto description: Service tier hint for request scheduling. store: type: boolean default: false description: >- Ignored — always forced to `false` because responses are not stored server-side. stream: type: boolean default: false description: >- If true, stream partial response events as Server-Sent Events (SSE). temperature: type: number nullable: true description: Sampling temperature; higher values increase randomness. text: type: object additionalProperties: true nullable: true description: >- Text output configuration, including structured output format via `text.format`. tool_choice: oneOf: - type: string - type: object additionalProperties: true default: auto nullable: true description: >- Controls which tool the model calls. Accepts a string (for example, `auto`, `none`, `required`) or an object specifying a function. tools: type: array items: type: object additionalProperties: true nullable: true description: Optional list of tool definitions the model may call. top_logprobs: type: integer default: 0 nullable: true description: >- Return the top N token logprobs at each position when logprobs are enabled. top_p: type: number nullable: true description: Nucleus sampling cumulative probability threshold. top_k: type: integer nullable: true description: >- Keep only the top_k tokens by probability at each generation step. truncation: type: string enum: - auto - disabled default: disabled description: >- Truncation strategy for inputs that exceed the model context window. skip_special_tokens: type: boolean default: true description: Whether to skip special tokens in the generated output. presence_penalty: type: number nullable: true description: Penalty applied to discourage new token presence. frequency_penalty: type: number nullable: true description: Penalty applied based on token frequency to reduce repetition. cache_salt: type: string nullable: true description: Optional salt for prefix caching. repetition_penalty: type: number nullable: true description: Additional repetition control for generation. seed: type: integer nullable: true description: Optional random seed for reproducibility. stop: oneOf: - type: string - type: array items: type: string nullable: true description: Stop sequence or list of sequences where generation should halt. structured_outputs: type: object additionalProperties: true nullable: true description: Structured output schema configuration. chat_template_kwargs: type: object additionalProperties: true nullable: true description: Additional keyword arguments passed to the model chat template. include: type: array items: type: string nullable: true description: Additional output fields to include in the response. previous_input_messages: type: array items: type: object additionalProperties: true nullable: true description: >- Prior conversation messages to include when continuing a multi-turn exchange without server-side response storage. previous_response_id: type: string nullable: true description: >- Not supported. Requests that set this field are rejected with a validation error because responses are not stored. background: type: boolean default: false description: >- Not supported. Requests with `background: true` are rejected with a validation error because responses are not stored. required: - model - input ResponsesResponse: type: object description: OpenAI-compatible response object returned by the Responses API. properties: id: type: string description: Unique identifier for this response. object: type: string default: response description: Object type. Always `response`. created_at: type: integer description: Unix timestamp (seconds) when the response was created. model: type: string description: Model identifier used to generate the response. status: type: string enum: - completed - failed - in_progress - incomplete - cancelled description: Status of the response generation. output: type: array items: $ref: '#/components/schemas/ResponseOutputItem' description: >- Typed output items produced by the model (for example, messages, function calls, or reasoning items). output_text: type: string nullable: true description: >- Convenience field containing concatenated text output from the response. instructions: type: string nullable: true description: System or developer instructions used for this response. usage: $ref: '#/components/schemas/ChatUsage' nullable: true description: Token usage details for the request. error: type: object nullable: true properties: code: type: string description: Error code. message: type: string description: Human-readable error description. description: Error details when generation fails. incomplete_details: type: object nullable: true properties: reason: type: string enum: - max_output_tokens - content_filter description: Why the response is incomplete. description: Details about why the response ended before completion. required: - id - object - created_at - model - status - output ResponseOutputItem: type: object description: >- A typed item in the response `output` array (for example, a message, function call, or reasoning item). properties: id: type: string description: Unique identifier for this output item. type: type: string description: >- Item type discriminator (for example, `message`, `function_call`, `reasoning`). role: type: string description: Role for message output items (for example, `assistant`). status: type: string enum: - in_progress - completed - incomplete description: Status of this output item. content: type: array items: type: object additionalProperties: true description: Content parts for message output items. call_id: type: string description: Identifier for function tool calls. name: type: string description: Function name for function call output items. arguments: type: string description: JSON-encoded arguments for function call output items. additionalProperties: true ChatUsage: type: object description: Token usage counts for chat completions. properties: prompt_tokens: type: integer description: Number of tokens consumed by the input messages. completion_tokens: type: integer description: Number of tokens generated by the assistant in the response. total_tokens: type: integer description: Total tokens consumed for the chat request. ````