# NVIDIA Sana Sprint AI Image Generator

> NVIDIA’s one/few-step diffusion model for extremely fast, inexpensive high-resolution drafts.

Canonical URL: https://allimageai.com/models/sana-sprint/

Website: All Image AI

Language: English

Sana Sprint is a one/few-step diffusion model from NVIDIA’s Sana project. It is designed for very low latency, and the hosted route supports custom dimensions plus named visual styles at a fraction of the cost of large final-render models.

## Model details

- Developer: NVIDIA
- Status on All Image AI: Active
- Last verified: 2026-08-25
- Maximum output: 4K
- Modes: Text to Image
- Reference images: not supported
- Resolutions: 1K, 2K, 4K
- Aspect ratios: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3

## Best for

- Very fast drafts
- High-volume prompt exploration
- Stylised concepts
- Low-cost high-resolution tests

## How to use

1. Start at 1K and two steps.
2. Choose a style only when the prompt does not define one.
3. Generate a broad batch.
4. Render 2K or 4K only after selecting a direction.

## Settings explained

### Steps

The operation defaults to two; Sprint is designed for very few steps.

### Style

Applies one of the endpoint’s documented visual presets without rewriting the prompt.

## Prompt examples

### Fast concept

> Compact desert research station at sunrise, solar awnings, dusty orange atmosphere, grounded industrial design

### Style study

> Tiny night market beneath giant glowing mushrooms, lively crowd, layered depth, whimsical fantasy illustration

## Limitations

- Text-to-image only
- Fine detail trails larger final-render models
- 4K should be reserved for prompts already proven at 1K

## Pricing and credits

Credits for one image at default quality and mode. Resolution, quality, reference inputs, edit mode and output count can change the total. Check the Generate button before submitting.

| Resolution | Credits / image |
| --- | --- |
| 1K | 1 |
| 2K | 1 |
| 4K | 3 |

Failed generations cost 0 credits. Catalog availability does not guarantee runtime availability; the workspace reports whether the selected model can run.

[Credit plans](https://allimageai.com/pricing.md)

## Test results

Testing in progress. No verified benchmark scores have been published for this model.

## Commercial use and image handling

Sana Sprint is released under Apache 2.0; generated use remains subject to applicable law and service terms.

Uploads are processed by the configured inference service. You can delete a generation from your history.

## Frequently asked questions

### Is Sana Sprint really a one-step model?

The research model is built around one/few-step generation; this endpoint defaults to two and allows up to twenty.

### Why does 4K cost two credits?

Provider billing follows output megapixels, so the 4K tier needs a separate margin-safe credit level.

## Release history and sources

March 22, 2025 — Open weights released

This date applies to the event shown. Later API access, weights and availability on All Image AI may have different dates.

- [github.com](https://github.com/NVlabs/Sana)

[Event evidence on the timeline](https://allimageai.com/timeline/#event-sana-sprint)

## Other versions

- [Sana 1.5 4.8B](https://allimageai.com/models/sana-1-5-4-8b.md): 4.8B · up to 4K
- [Sana 1.5 1.6B](https://allimageai.com/models/sana-1-5-1-6b.md): 1.6B · up to 4K