sana-sprint — AI Image Model

nvidia

nvidia

sana-sprint

Commercial useimage

SANA-Sprint 1.6B is best for ultra-fast, high-quality text-to-image generation, producing 1024×1024 images in just 0.1 seconds. The model is a One-Step Diffusion with Continuous-Time Consistency Distillation developed by NVIDIA, designed to handle various natural language processing tasks with high efficiency which also includes multiple specialized endpoints, each optimized for a particular type of task. This allows users to achieve better results by choosing the most suitable endpoint for their needs.

Model Type
coin

Pricing: 4.6 joules / generation

Input
inference steps
Number of sampling steps
14
intermediate timesteps
Intermediate timestep value (only used when inference_steps=2, recommended values: 1.0-1.4)
0.51.5
guidance scale
CFG guidance scale
1.020.0
output quality
Quality when saving the output images, from 0 to 100. 100 is best quality, 0 is lowest quality. Not relevant for .png outputs
0100
OutputIMAGE

Output will be displayed here

Fill in the inputs on the left and hit RUN to preview the results.

Examples

Explore different use cases and parameter configurations

example
image