---
title: OpenAI | Scenario Docs
---

> This page is auto-generated from model configurations. Last updated: 2026-09-25.

This reference lists all available **OpenAI** 3d generation models and their parameters. Use these parameter names when calling the [Generation API](/api/postgeneratecustom/index.md).

- [Scenario GPT-6 Astra 3D](#scenario-gpt-6-astra-3d)

---

## Scenario GPT-6 Astra 3D

Turn reference images into an editable 3D object — named parts, clean topology, per-part materials. Best for props and hard-surface subjects, not characters.

**Model ID:** `model_scenario-gpt6-astra-3d`

**Capabilities:** `img23d`

**LLM Markdown:** <https://app.scenario.com/api/models/model_scenario-gpt6-astra-3d/markdown>

| Parameter     | Type        | Required | Default   | Min  | Max | Allowed Values                | Description                                                                                                                                                             |
| ------------- | ----------- | -------- | --------- | ---- | --- | ----------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `images`      | file\_array | Yes      | -         | -    | -   | -                             | 1 to 8 photos or renders of one object. The first image sets the viewpoint the result is checked against; extra views help with the unseen sides.                       |
| `prompt`      | string      | No       | -         | -    | -   | -                             | Name the object in a few words, e.g. ‘a tugboat’. Worth filling in when the reference has clutter or more than one object in it, so the model knows which one to build. |
| `buildEffort` | string      | No       | `maximum` | -    | -   | `standard`, `high`, `maximum` | How much work goes into modeling the object. Use High or Maximum for mechanical or detailed subjects; Standard is faster and cheaper for simple shapes.                 |
| `refineSteps` | number      | No       | `7`       | 0    | 10  | -                             | Maximum correction passes. The run stops on its own once the result stops improving; a lower cap saves cost but can cut it off before that point.                       |
| `faceBudget`  | number      | No       | `250000`  | 1000 | -   | -                             | Hard cap on the delivered model’s triangle count, applied at export. 10k–50k for game-ready assets; 250k keeps fine detail.                                             |
