RLM

Recursive Language Model endpoints.

Create an RLM response

post

OpenAI Responses-style endpoint for Recursive Language Model execution. Accepts a strict subset of the Responses request shape and proxies to the internal RLM backend.

Authorizations

ApiKeyAuth

AuthorizationstringRequired

API key in Authorization header using Bearer .

Body

application/json

Strict OpenAI Responses-style request supported by the RLM gateway endpoint.

modelstringRequired

Model slug to run through the RLM backend.

inputany ofRequired

String input or a text-only array of Responses message items.

stringOptional

or

RLMResponsesInputMessage[]Optional

instructionsstring · nullableOptional

Optional top-level instructions mapped to the backend root prompt.

backgroundboolean · nullableOptional

When true, start the RLM run in the background and return an in-progress response resource that can be retrieved later.

Default: false

reasoningany of · nullableOptional

Optional reasoning controls.

storeboolean · nullableOptional

Must be false or omitted. Stateful Responses storage is not supported.

streamboolean · nullableOptional

Must be false or omitted. Streaming is not supported for this route.

rlmany of · nullableOptional

RLM-specific execution controls.

Responses

200

Successful Response

application/json

idstringRequired

objectconst: responseOptionalDefault: response

created_atintegerRequired

statusstring · enumRequiredPossible values:

modelstringRequired

storebooleanOptionalDefault: false

output_textstring · nullableOptional

400

Bad Request

application/json

401

Unauthorized

application/json

402

Payment Required

application/json

403

Forbidden

application/json

404

Not Found

application/json

409

Conflict

application/json

429

Too Many Requests

application/json

500

Internal Server Error

application/json

501

Not Implemented

application/json

503

Upstream Unavailable

application/json

post/v1/rlm/responses

HTTPcURLJavaScriptPython

POST /v1/rlm/responses HTTP/1.1
Host: platform.maniac.ai
Authorization: Bearer YOUR_SECRET_TOKEN
Content-Type: application/json
Accept: */*
Content-Length: 257

{
  "background": false,
  "input": "Summarize this document.",
  "instructions": "Be concise and factual.",
  "metadata": {
    "source": "dashboard"
  },
  "model": "openai/gpt-4o",
  "reasoning": {
    "effort": "medium"
  },
  "rlm": {
    "max_depth": 2,
    "max_iterations": 30
  },
  "store": false,
  "stream": false
}

{
  "created_at": 1710000000,
  "id": "resp_123",
  "metadata": {
    "source": "dashboard"
  },
  "model": "openai/gpt-4o",
  "object": "response",
  "output": [
    {
      "content": [
        {
          "annotations": [],
          "text": "Summary goes here.",
          "type": "output_text"
        }
      ],
      "id": "msg_123",
      "role": "assistant",
      "status": "completed",
      "type": "message"
    }
  ],
  "output_text": "Summary goes here.",
  "status": "completed",
  "store": false,
  "usage": {
    "input_tokens": 120,
    "output_tokens": 30,
    "total_tokens": 150
  }
}

Get an RLM response

get

Retrieve a background RLM response resource by id within the authenticated project.

Authorizations

ApiKeyAuth

AuthorizationstringRequired

API key in Authorization header using Bearer .

Path parameters

response_idstringRequired

Responses

200

Successful Response

application/json

idstringRequired

objectconst: responseOptionalDefault: response

created_atintegerRequired

statusstring · enumRequiredPossible values:

modelstringRequired

storebooleanOptionalDefault: false

output_textstring · nullableOptional

400

Bad Request

application/json

401

Unauthorized

application/json

403

Forbidden

application/json

404

Not Found

application/json

409

Conflict

application/json

422

Validation Error

application/json

429

Too Many Requests

application/json

500

Internal Server Error

application/json

501

Not Implemented

application/json

503

Upstream Unavailable

application/json

get/v1/rlm/responses/{response_id}

HTTPcURLJavaScriptPython

GET /v1/rlm/responses/{response_id} HTTP/1.1
Host: platform.maniac.ai
Authorization: Bearer YOUR_SECRET_TOKEN
Accept: */*

{
  "created_at": 1710000000,
  "id": "resp_123",
  "metadata": {
    "source": "dashboard"
  },
  "model": "openai/gpt-4o",
  "object": "response",
  "output": [
    {
      "content": [
        {
          "annotations": [],
          "text": "Summary goes here.",
          "type": "output_text"
        }
      ],
      "id": "msg_123",
      "role": "assistant",
      "status": "completed",
      "type": "message"
    }
  ],
  "output_text": "Summary goes here.",
  "status": "completed",
  "store": false,
  "usage": {
    "input_tokens": 120,
    "output_tokens": 30,
    "total_tokens": 150
  }
}