FLUX 3 Image Klein AI
FLUX 3 Image Klein is Black Forest Labs' compact 5.6B open-weights FLUX 3 Image model: fast 16-step text-to-image in 15 aspect ratios up to 4K, and editing with up to 10 reference images. Released October 2026. Try it online in the Studio or call it from code, up to 87% below the official price.
- Developer
- Black Forest Labs
- FLUX family
- Released
- October 2026
- Sizes
- up to 4K
- 5 sizes
- From
- $0.0011/image
- 768
Try it
Try FLUX 3 Image Klein AI now
Runs on your balance at the price on the button; failed runs are refunded. Results are saved to your account.
What it is
What is FLUX 3 Image Klein?
FLUX 3, distilled
Klein is the small FLUX 3 Image model — 5.6B parameters, step-distilled to 16 steps — so images come back in seconds with FLUX 3's look for photos, illustration and design, at a fraction of FLUX 3's price.
Write the whole scene
Text-to-image up to 2K runs Black Forest Labs' distilled 16-step Klein weights — the checkpoint BFL's own API serves — on optimized GPU inference for open-source models. Every request, editing and 4K included, goes in without the upsampler BFL's API puts in front of Klein, which rewrites short prompts into full captions. Klein is trained on such captions: describe the subject, setting, light and framing yourself. A one-line prompt gives a plainer image than BFL's API would.
Edit with references
Give it up to 10 images and describe the change — "put the lamp from image 2 on the desk in image 1" — one price per image however many references, below BFL's own.
Capabilities
What FLUX 3 Image Klein can do
Fraction of a cent
Per 1K image
16 steps
Seconds per image
Up to 4K
768 to 4K
10 references
Image editing
Examples
FLUX 3 Image Klein examples
Prompts you can load into the Studio and try.
Editorial still life of a hand-thrown terracotta vase with dried orange ranunculus on a travertine plinth, low golden-hour sun from the left, long soft shadows on a warm plaster wall, medium-format film photograph
Wide shot of a night market in Taipei: steam rising from a dumpling stall as a cook lifts a bamboo steamer lid, neon signs reflected on the wet pavement, people browsing behind, handheld 35mm film photo with fine grain
Art deco travel poster of a streamlined train crossing a coastal stone viaduct at sunset, bold gold lettering reading "RIVIERA EXPRESS" across the top, navy and gold geometric sunburst behind the train
Use cases
What people use FLUX 3 Image Klein for
- Images from a description
- Illustrations, product visuals and concepts, up to 4K.
- Editing your images
- Change backgrounds, colours or outfits, or combine up to 10 images in one edit.
- Any format
- 16 aspect ratios, from banners to vertical posts.
Versions
FLUX 3 Image Klein in the FLUX family
Every FLUX model we offer, newest first. FLUX AI covers the whole family.
| Model | Released | Max size | From |
|---|---|---|---|
| FLUX 3 Image Klein | October 2026 | 4K | $0.0011/image |
| FLUX 3 | October 2026 | 4K | $0.0317/image |
| FLUX 3 Video | August 2026 | 720p · 5–20 s | $0.1564/s |
| FLUX.2 [klein] 4B | January 2026 | 2K | $0.001/image |
| FLUX.2 [max] | December 2025 | 4 MP | $0.0644/image |
| FLUX.2 [pro] | November 2025 | 4 MP | $0.0276/image |
| FLUX.2 [flex] | November 2025 | 4 MP | $0.046/image |
| FLUX.1 Kontext [pro] | May 2025 | 1K | $0.0348/image |
| FLUX.1 Kontext [max] | May 2025 | 1K | $0.0695/image |
How it compares
FLUX 3 Image Klein vs similar models
FLUX 3 Image Klein vs FLUX 3
FLUX 3 Image Klein
- Lower price: $0.0019/image vs $0.0317/image at 1K (94% less).
- More aspect ratios: 16 vs 15.
- Newer model (released October 15, 2026).
FLUX 3
- A different balance of price and features.
Where to use it
Three ways to use FLUX 3 Image Klein
In the Studio
No code: write a prompt, pick the settings, generate. The price is shown before you run.
Through the API
One key and one request shape for every model; poll the generation or get a webhook.
FAQ
FLUX 3 Image Klein AI questions
FLUX 3 Image Klein is Black Forest Labs' open-weights model. CheaperGen provides API access to the open weights, run on GPU inference for open-source models, and does not own the model. Trademarks belong to their respective owners.