← Back to all generators

nvidia/sana

A fast image model with wide artistic range and resolutions up to 4096x4096

Capabilities

1:1 4:3 3:4 16:9 9:16 2:3 3:2 Negative Prompt Seed

Cost

Community model (estimated from hardware time)

Input Parameters

guidance_scale number

Classifier-free guidance scale

Default: 5 min: 1, max: 20
height integer

Height of output image

Default: 1024
model_variant string

Model variant. 1600M variants are slower but produce higher quality than 600M, 1024px variants are optimized for 1024x1024px images, 512px variants are optimized for 512x512px images, 'multilang' variants can be prompted in both English and Chinese

Default: "1600M-1024px"
1600M-1024px 1600M-1024px-multilang 1600M-512px 600M-1024px-multilang 600M-512px-multilang
negative_prompt string

Specify things to not see in the output

Default: ""
num_inference_steps integer

Number of denoising steps

Default: 18 min: 1
pag_guidance_scale number

PAG Guidance scale

Default: 2 min: 1, max: 20
prompt string

Input prompt

Default: "a cyberpunk cat with a neon sign that says "Sana""
seed integer

Random seed. Leave blank to randomize the seed

width integer

Width of output image

Default: 1024
Version: c6b5d2b74599 Updated: 2/26/2026 238.9K runs