23 July 2026: FLUX 3 announced
Black Forest Labs announced FLUX 3 as one model family for images, video, audio and robot action. Video and Action were opened in early access, with FLUX 3 Image to follow. Source: press coverage of the announcement.
FLUX 3
FLUX 3 is Black Forest Labs' multimodal model family for image, video, audio and robot action. Currently generating with Nano Banana 2.1, not FLUX · we plan to add a FLUX 3 image model here once it is available to us. The generator below is Nano Banana 2.1, so its images are not FLUX images.
Currently generating with Nano Banana 2.1, not FLUX · we plan to add a FLUX 3 image model here once it is available to us.
The generator on this page is the same one as the Nano Banana 2.1 page. For a newer OpenAI image model, see GPT Image 2.5.
01
The generator on this page runs Nano Banana 2.1. VibeAha does not offer FLUX 3 or FLUX.2.
02
Describe subject, framing, lighting and format. Add up to 10 reference images when the subject must stay consistent, then pick ratio and resolution.
03
Submit, check subject and composition at full size, then change one instruction at a time.
Black Forest Labs announced FLUX 3 as one model family for images, video, audio and robot action. Video and Action were opened in early access, with FLUX 3 Image to follow. Source: press coverage of the announcement.
Reported capabilities are clips of up to 20 seconds with synchronized audio, from text, an image, a reference video or keyframes.
Black Forest Labs released FLUX 3 Image Klein, a 5.6 billion parameter image model, as open weights and through its own API. Source: Black Forest Labs launch post and press coverage.
No FLUX model is connected to VibeAha, so this page generates with Nano Banana 2.1.
Stated by Black Forest Labs
FLUX 3 covers image, video, audio and robot action prediction in one model family, according to Black Forest Labs.
It handles text to image, image editing and multi-reference generation with up to 10 reference images, with a 5.6B parameter size.
FLUX.2 [klein] is an earlier, separate release in 4B and 9B sizes. FLUX 3 Image Klein is the newer 5.6B model.
Reported, unclear or not available here
Press reports say commercial use is free for individuals and companies under $5 million annual revenue under the FLUX Kommunity License. Read the license text before you rely on it.
Sources differ on whether Klein is the whole FLUX 3 image release or a smaller part of it. Pricing for FLUX 3 Video and Action is TBA.
This page runs Nano Banana 2.1. It cannot make FLUX 3 video, audio or Klein images.
| FLUX 3 | FLUX.2 | |
|---|---|---|
| Modalities | Image, video, audio and robot action | Image generation and editing |
| Smaller image model | FLUX 3 Image Klein, 5.6B, open weights (15 October 2026) | FLUX.2 [klein], 4B and 9B |
| Video | Up to 20 seconds in early access, reported | Not part of FLUX.2 |
| Other FLUX 3 image sizes | TBA | Not applicable |
| On VibeAha | Not offered. This page runs Nano Banana 2.1 | Not offered. This page runs Nano Banana 2.1 |
| Nano Banana 2.1 credits at 1K / 2K / 4K | 3 / 5 / 8 | 3 / 5 / 8 |
FLUX entries come from Black Forest Labs or press coverage of it, or are marked TBA. The last row is the credit quote for the generator on this page. VibeAha does not publish benchmarks between FLUX and Nano Banana.
| Option | Best for | Key difference |
|---|---|---|
| Nano Banana 2.1 | The generator on this page | Up to 10 reference images and up to 4K output. |
| GPT Image 2.5 | Creating and editing images with the latest GPT Image model | Offered as two workspaces, Flare and Sunburst. |
| Grok Imagine Image 2 | Region level edits | Creates images, maps regions and edits a selection. |
Model capabilities were verified on 2026-10-16. Available settings and credits are checked again when you submit.