overlay: 1.0.0
info:
  title: API Evangelist enhancements for the RouterPlex Inference API
  version: 1.0.0
extends: openapi/routerplex-inference-openapi.json
x-generated: '2026-08-09'
x-method: generated
x-source: >-
  Derived from the provider's own published surface: docs.routerplex.com
  (llms-full.txt), routerplex.com/pricing.md, /.well-known/api-catalog, and the
  live MCP server at routerplex.com/mcp. Never mutates the harvested spec.
actions:
  - target: $.info
    update:
      x-apievangelist-enriched: '2026-08-09'
      x-api-catalog: https://routerplex.com/.well-known/api-catalog
      x-llms-txt: https://routerplex.com/llms.txt
      x-mcp-server: https://routerplex.com/mcp
      x-content-signal: 'ai-train=no, search=yes, ai-input=yes'
      x-launched: '2026-07'

  # The spec omits the Anthropic-compatible x-api-key header that the docs document
  # as an equal-standing auth style on /messages.
  - target: $.components.securitySchemes
    update:
      RouterPlexAnthropicKey:
        type: apiKey
        in: header
        name: x-api-key
        description: >-
          Anthropic-compatible clients send the same RouterPlex API key in x-api-key.
          Documented at https://docs.routerplex.com/authentication — both styles work
          on /v1/messages.

  # Platform limits published at https://docs.routerplex.com/errors but absent from the spec.
  - target: $.info
    update:
      x-rate-limit:
        edge: 60 requests/second per IP
        request_body_max: 50 MB
        stream_max_duration: 10 minutes
        note: >-
          Separate from optional per-key RPM/TPM limits and the tighter limits on
          promotional accounts. Plan tiers additionally apply 4-hour, weekly and
          monthly spend windows (https://routerplex.com/pricing.md).

  - target: $.paths['/chat/completions'].post
    update:
      tags: [chat]
      x-billable: true
      x-streaming: server-sent-events

  - target: $.paths['/images/generations'].post
    update:
      tags: [images]
      x-billable: true

  - target: $.paths['/responses'].post
    update:
      tags: [chat]
      x-billable: true
      x-streaming: server-sent-events
      x-compatibility: OpenAI Responses API — intended for Codex custom model providers.

  - target: $.paths['/messages'].post
    update:
      tags: [chat]
      x-billable: true
      x-streaming: server-sent-events
      x-compatibility: >-
        Anthropic Messages API. Returns Anthropic SSE events (message_start,
        content_block_delta, message_stop) even when the selected model is not a
        Claude model.

  - target: $.paths['/models'].get
    update:
      tags: [models]
      x-billable: false
      x-mcp-equivalent: routerplex_list_models
      description: >-
        Returns the models available to the authenticated key, honoring that key's
        model allowlist. The full public catalog with prices is available
        anonymously via the MCP tool routerplex_list_models or at
        https://routerplex.com/models.

  # The spec declares no response schemas; the docs publish the OpenAI-shaped error envelope.
  - target: $.components
    update:
      schemas:
        RouterPlexError:
          type: object
          description: >-
            OpenAI-shaped error envelope, documented at
            https://docs.routerplex.com/errors. Applied to every 4xx/5xx response.
          properties:
            error:
              type: object
              properties:
                message: {type: string}
                type: {type: string, example: invalid_request_error}
                code: {type: string}

  - target: $.tags
    update:
      - name: chat
        description: OpenAI-, Responses- and Anthropic-compatible completion routes.
      - name: images
        description: OpenAI-compatible image generation.
      - name: models
        description: Model catalog available to the authenticated key.