What Is Stable Image Ultra?
If you already work the local Stable Diffusion way — negative prompts, CFG tuning, step counts — this is the lowest-friction move to a hosted model. The classic parameter set is intact and you no longer manage checkpoints or a GPU.
Stable Diffusion Model Versions
Stable Image Ultra
Recommended · 2K
Coming soon
Stable Diffusion 3.5 Large
The full 8B SD 3.5 model with negative prompts, CFG and sampling-step control.
- Resolutions: 1K / 2K
- Reference images: not supported
- Modes: Text to Image
- Speed: Balanced
Stable Diffusion 3.5 Large Turbo
The four-step distilled SD 3.5 Large for fast, controllable open-model drafts.
- Resolutions: 1K / 2K
- Reference images: not supported
- Modes: Text to Image
- Speed: Fastest
Stable Diffusion 3.5 Medium
The resource-efficient SD3.5 tier with classic CFG, steps and negative-prompt controls.
- Resolutions: 1K / 2K
- Reference images: not supported
- Modes: Text to Image, Typography
- Speed: Balanced
Stable Diffusion 3 Medium
Stability AI’s two-billion-parameter SD3 checkpoint with typography, negative prompts and familiar diffusion controls.
- Resolutions: 1K
- Reference images: not supported
- Modes: Text to Image, Typography
- Speed: Balanced
Stable Image Ultra
The full classic parameter set: Negative Prompt, Steps, Guidance, Strength.
- Resolutions: 1K / 2K
- Reference images: up to 1
- Modes: Text to Image, Image to Image, Edit Image
- Speed: Balanced
What Is Stable Image Ultra Best For?
- People fluent in the classic parameters
- Work that needs negative prompts to exclude elements
- Fine-grained image-to-image control
- Stylised illustration
How to Use Stable Image Ultra
- Select Stable Image Ultra.
- Write the positive prompt, then open Advanced for the negative prompt.
- Fix the seed and change one parameter at a time.
- Upscale to the delivery size.
Stable Image Ultra Settings Explained
Guidance
The classic CFG. 4 is a reasonable starting point.
Steps
30 covers most cases; returns fall off quickly after that.
Negative Prompt
Excluding unwanted elements is one of this model’s most effective controls.
The panel above shows only the parameters Stable Image Ultra actually supports. Switch versions and unsupported controls disappear rather than sitting there doing nothing.
Stable Image Ultra Prompt Examples
With a negative prompt
Fantasy forest path at dusk, volumetric light, painterly
Negative Prompt: blurry, watermark, text.
Image to image
Repaint this sketch as a finished digital illustration, keep the composition
Set Strength around 0.5.
Stable Image Ultra 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 Stable Image Ultra, so this page shows no scores. We do not publish numbers we have not measured.
Stable Image Ultra Limitations
- 2K ceiling
- Weak in-image text
- Less realistic by default than FLUX.2
Stable Image Ultra Pricing and Credits
Credits scale with resolution and image count. The Generate button always shows what this run costs.
1K
2 credits / image
2K
3 credits / image
Failed generations cost 0 credits.
Commercial Use and Image Rights
Commercial use is subject to Stability AI's licence terms.
- Model used: Stable Image Ultra
- Model developer: Stability AI
- 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
Do the parameters match local SD?+
Conceptually yes: Guidance maps to CFG, Steps to sampling steps, Strength to denoising strength.
Is LoRA supported?+
Custom LoRAs are not exposed in this integration.
Release History and Sources
June 5, 2024 · API access opened
This date applies to the event shown. Later API access, weights and availability on All Image AI may have different dates.
Explore this event on the image model timeline →
Related Models and Tools
AI Image Generator
Create an image, try a new look or transform a photo. Start with the result you want.
Image to Image
Start from an image. You decide how much of it survives.
Image Upscaler
Scale up to print size and rebuild texture on the way.
FLUX.2 Max
Black Forest Labs
The highest-fidelity FLUX.2 tier for final-publish generation and demanding prompt adherence.
FLUX.2 Flex
Black Forest Labs
FLUX.2 with explicit guidance and step controls for typography-heavy final renders.
FLUX.2 Flash
Black Forest Labs
A low-cost FLUX.2 route for rapid generation and exact multi-reference edits.
FLUX.2 Turbo
Black Forest Labs
A fast FLUX.2 draft model with prompt expansion and unusually low megapixel cost.