Oxagen Docs

Model Selection

How to choose which model powers a Content Studio generation — the per-generation model selector in the chat composer.

Content Studio generation runs use the workspace's configured model tier by default. You can override this on a per-message basis using the model picker in the chat composer.

The model picker

The model picker appears as a pill badge in the bottom-left corner of the chat composer. It shows the currently active selection. Clicking the pill opens a popover with two groups:

Workspace tiers — the three Oxagen-managed tiers (Oxagen, Oxagen+, Oxagen+ Pro). Each tier maps to a model chosen in workspace settings. Tiers that require a Pro plan show a lock icon when the workspace is not on Pro.

My models — any bring-your-own-key models the workspace has added via Settings → Models → API Keys. These appear only when at least one provider key is connected.

A Configure models link at the bottom of the popover navigates to Settings → Models to manage tiers and keys.

Changing the model for a generation

  1. Open Content Studio in the workspace sidebar.
  2. Type or paste a generation prompt in the composer.
  3. Click the model pill to open the picker.
  4. Select a tier or a specific BYOK model.
  5. Submit the prompt. The generation run uses the selected model for script generation.

The selection persists for the duration of the conversation. Starting a new conversation resets the selection to the workspace default.

What the model selector controls

The model selector affects the script generation step — the LLM call that turns your prompt and workspace graph data into a structured scene plan. It does not control the image, audio, or video rendering providers; those are set workspace-wide in Workspace Defaults.

Selector valueScript generation model
Workspace default (no selection)The model assigned to the workspace's Level 1 tier
Oxagen tierLevel 1 model for the workspace
Oxagen+ tierLevel 2 model for the workspace (Pro required)
Oxagen+ Pro tierLevel 3 model for the workspace (Pro required)
BYOK model IDThe specific model from the connected provider

API: passing a model selector

When creating a generation run via the REST API, pass model_selector on the POST /v1/workspaces/{workspace_id}/content-studio/runs body:

{
  "prompt": "10-minute onboarding video for new engineers",
  "output_types": ["video"],
  "target_duration_seconds": 600,
  "template_id": "onboarding",
  "visibility": "workspace",
  "model_selector": "level_2"
}

model_selector accepts either a tier string ("level_1", "level_2", "level_3") or a model ID from the workspace catalog (e.g. "gpt-5-mini"). Omit the field to use the workspace default.

Invalid values return 422 Unprocessable Entity.

On this page