What Is Sana 1.5 4.8B?
Sana 1.5 4.8B is NVIDIA’s larger inference-scaling checkpoint. It improves quality and prompt alignment over the compact tier while retaining the Sana architecture’s efficient high-resolution design and classic diffusion controls.
Sana Image Model Versions
Sana 1.5 4.8B
4.8B · up to 4K
From 1 Credits
Sana 1.5 4.8B
NVIDIA’s larger Sana 1.5 checkpoint for stronger detail while retaining efficient high-resolution generation.
- Resolutions: 1K / 2K / 4K
- Reference images: not supported
- Modes: Text to Image
- Speed: Balanced
Sana 1.5 1.6B
NVIDIA’s lightweight Sana 1.5 checkpoint for efficient open-model generation all the way to 4K.
- Resolutions: 1K / 2K / 4K
- Reference images: not supported
- Modes: Text to Image
- Speed: Fast
Sana Sprint
NVIDIA’s one/few-step diffusion model for extremely fast, inexpensive high-resolution drafts.
- Resolutions: 1K / 2K / 4K
- Reference images: not supported
- Modes: Text to Image
- Speed: Fastest
What Is Sana 1.5 4.8B Best For?
- Detailed open-model finals
- High-resolution illustration
- Prompt-aligned scenes
- 4K experimentation
How to Use Sana 1.5 4.8B
- Write the complete visual brief.
- Start at 1K and eighteen steps.
- Compare fixed seeds or style presets.
- Render only the winner at 2K or 4K.
Sana 1.5 4.8B Settings Explained
Style
Uses one of Sana’s documented presets; leave it unset when the prompt already defines a medium.
Resolution
Iterate at 1K because provider cost follows output megapixels; reserve 4K for the chosen seed.
The panel above shows only the parameters Sana 1.5 4.8B actually supports. Switch versions and unsupported controls disappear rather than sitting there doing nothing.
Sana 1.5 4.8B Prompt Examples
High-resolution landscape
Layered volcanic coast at sunrise, black sand, sea mist, tiny research cabin for scale, richly textured cinematic matte painting
Styled object
Mechanical songbird made of brass and blue enamel, perched on a simple branch, precise product illustration, clean negative space
Sana 1.5 4.8B Test Results
Testing in progress
Every model runs the same eight-category prompt set — portrait, product, poster text, anime character, interior, multi-reference, image editing and character consistency — recording generation time, success rate, text accuracy and identity retention. That run has not finished for Sana 1.5 4.8B, so this page shows no scores. We do not publish numbers we have not measured.
Sana 1.5 4.8B Limitations
- Slower than the 1.6B checkpoint
- Text-to-image only
- Develop the prompt at 1K before paying for 4K
Sana 1.5 4.8B Pricing and Credits
Credits scale with resolution and image count. The Generate button always shows what this run costs.
1K
1 credits / image
2K
2 credits / image
4K
8 credits / image
Failed generations cost 0 credits.
Commercial Use and Image Rights
Sana 1.5 is released under Apache 2.0; generated use remains subject to applicable law and service terms.
- Model used: Sana 1.5 4.8B
- Model developer: NVIDIA
- The inference route is managed internally and may change without changing the selected model
- Uploads are sent only to the configured inference service needed to complete the generation
- You can delete any generation from your history at any time
Frequently Asked Questions
Should I choose 4.8B or 1.6B?+
Choose 4.8B for final detail and 1.6B for faster, less expensive exploration.
Is Sana 1.5 open source?+
The official NVIDIA repository is Apache 2.0 and publishes the Sana 1.5 training and inference stack.
Model Updates and Changelog
- 2026-08-25 —— Added the exact Sana 1.5 4.8B route with 1K, 2K and 4K pixel pricing.
Related Models and Tools
AI Image Generator
One prompt box, every major model. Describe it, pick a model, generate.
Social Post Generator
One visual, every platform size, one consistent style.
FLUX.2 Max
Text To Image · Typography
FLUX.2 Flex
Text To Image · Typography
FLUX.2 Flash
Text To Image · Image Editing · Multi Reference
FLUX.2 Turbo
Text To Image · Image To Image · Image Editing