GET
/v1/schemas/meta/chat/v1%2Fresponses
Accept:
Request body
28 properties · 0 required · 87 $defs| property | type | description | constraints |
|---|---|---|---|
| metadata | object | nullable | |
| top_logprobs | integer | An integer between 0 and 20 giving the number of top alternative tokens to return at each position, each with its log probability. (minimum: 0, maximum: 20) | 0 ≤ n ≤ 20 |
| temperature | number | What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. We generally recommend altering this or top_p but not both. (minimum: 0, maximum: 2, default: 1) | nullable |
| top_p | number | Nucleus sampling: only tokens whose cumulative probability is at most this value are considered. Must be greater than 0; values <= 0 return HTTP 400. Set this or temperature, not both. (minimum: 0, maximum: 1, default: 1) | nullable |
| user | string | Deprecated. Being replaced by safety_identifier and prompt_cache_key; use prompt_cache_key to retain caching benefits. A stable end-user identifier used to improve cache hit rates and help detect abuse. | |
| safety_identifier | string | A stable identifier that helps detect users who may be violating usage policies. Use a string that uniquely identifies each user, up to 64 characters; hash a username or email so you do not send identifying information. (maxLength: 64) | length ≤ 64 |
| prompt_cache_key | string | Groups similar requests to improve cache hit rates. Replaces the user field. | |
| service_tier | enum | auto | default | flex | scale | priority · nullable | |
| prompt_cache_retention | enum | The retention policy for the prompt cache. | in_memory | 24h · nullable |
| previous_response_id | string | Identifier of a previous response to continue from. Use this to build multi-turn conversations with server-managed state. Cannot be combined with conversation. | nullable |
| model | string | ID of the model used to generate the response. | |
| reasoning | reasoning | nullable | |
| background | boolean | Whether to run the response asynchronously in the background. Cannot be combined with stream. (default: false) | nullable |
| max_tool_calls | integer | Maximum number of built-in tool calls allowed while generating a response, counted across all built-in tools rather than per tool. The model ignores any calls beyond this limit. (minimum: 1) | nullable |
| text | response-text-param | ||
| tools | tool[] | ||
| tool_choice | tool-choice-param | ||
| prompt | object | nullable | |
| truncation | enum | Truncation strategy for the request. Only disabled (the default) is supported: if the input plus requested output tokens exceed the model's context window, the request fails with a 400 error. The auto strategy is not supported. (default: disabled) | auto | disabled · nullable |
| client_metadata | object | Optional client-provided request metadata. Accepted for compatibility and ignored. | nullable |
| input | string | input-item[] | ||
| include | include-enum[] | Additional output data to return in the response. Use message.output_text.logprobs to return token-level log probabilities on output text (set together with top_logprobs and a reasoning.effort of none), and reasoning.encrypted_content to return encrypted reasoning items for stateless multi-turn replay. | nullable |
| parallel_tool_calls | boolean | Whether the model may issue multiple tool calls in a single turn. (default: true) | nullable |
| store | boolean | Whether to persist the generated response so it can be retrieved later and chained from previous_response_id. (default: true) | nullable |
| instructions | string | Developer-level instructions inserted at the top of the model's context. When previous_response_id is set, instructions from the prior response are not carried forward, so you can change them on each turn. Top-level image models place these instructions immediately before the current user message instead, and also honor developer or system input messages that come before the first user message or after the most recent image_generation_call; those from earlier turns, and those stored with a previous response, are not carried forward. Top-level image models accept at most 12,000 characters across this field and those messages, returning a 400 error when the combined limit is exceeded. | nullable |
| stream | boolean | If true, the response is streamed to the client as server-sent events as it is generated. Cannot be combined with background. (default: false) | nullable |
| stream_options | object | nullable | |
| max_output_tokens | integer | Upper bound on the number of tokens generated for the response, counting both visible output and reasoning tokens. Minimum is 16; values above the model's configured maximum, or that exceed the remaining context window, return a 400 error. (minimum: 16) | nullable |
Validate a payload
before you spend tokensshell
$ curl -X POST https://modelschemas.com/v1/validate \
-d '{"provider":"meta","endpointId":"v1/responses","payload":{…}}'
→ {"valid": false, "errors": [{"path": "#", "keyword": "required", …}]}