Token360

Native Params Passthrough

Token360 supports two ways to call multimodal APIs:

  1. Normalized (default) — OpenAI-style fields (prompt, reference_images, seconds, …). The router maps them to each vendor’s API.
  2. Native params passthrough — You send the target model’s native JSON body. Enable it with the header below.

This document describes the second mode: the required header, supported endpoints, and common request issues.

Request header

Token360-Native-Paramstrue, 1, yes, on (case-insensitive)

When enabled, the HTTP body must be a JSON object (Content-Type: application/json). Invalid JSON (e.g. trailing commas), empty body, or a non-object root (array/string) will be rejected.

What Token360 still handles

Even in native mode, Token360 still:

  1. Authenticates your request with your Token360 API key.
  2. Uses the model value to select the requested model.
  3. Validates asset://... references against assets available to your account.
  4. Returns Token360-shaped task IDs and billing metadata where applicable.

Use Token360 asset IDs in asset://{assetId} references. Billing and polling remain on Token360 endpoints.

Endpoints that support this header

POST /v1/videosVideo generationJSON only; no multipart/form-data in native mode.
POST /v1/images/generationsImage generationBody must be a non-empty JSON object ({} is rejected).
POST /v1/audio/speechText-to-speechSkips the subset of OpenAI-style validation (input / voice required in default mode); body is passed through after model / asset:// handling.

Unsupported models return a clear not-supported error for native mode.

Common errors

JSON object body is required when Token360-Native-Params is enabledEmpty body, wrong Content-Type, invalid JSON (trailing comma), or JSON root not an object.
InvalidParameter on durationString "4" instead of number 4, or duration not in the model’s allowed set.
Asset validation failedasset:// id does not match an available Token360 asset, or typo in the id string.
  • Default OpenAI-style video doc in this portal: API Reference → Video Generation → Submit video generation request.
Was this page helpful?