Turn a photo, a first and last frame, or a text prompt into a 5–20 second Flux 3 Max video with sound, at 720p or 1080p. No API key and no setup: sign in, pick a mode, and generate.
Recent Generations
Flux 3 Max is a video generation model released by fal on October 19, 2026, post-trained with Black Forest Labs on their FLUX 3 foundation model. It turns a still image, a first and last frame, or a text prompt into a 5 to 20 second clip with generated sound. It is a video model, not an image model, and it is built for speed: fal reports about 12 seconds to render a 10-second clip.
Four steps take you from an idea to a downloadable MP4.
Image to Video animates one photo, Text to Video starts from words, and First & Last Frame moves from one image to another.
Upload your image or keyframes, then describe the motion, the camera, and the sound. Tap an example prompt to start faster.
Pick 5 to 20 seconds and 720p or 1080p. Turn on Preview mode to test a prompt for fewer credits first.
Press Generate video. Your clip is saved to your history, ready to download or run again with a new prompt.
Flux 3 Max trades a little length for a lot of speed. The figures below are the ones published at launch.
| Model | Time for a 10s clip | Image-to-video arena | Longest clip | Reference video input |
|---|---|---|---|---|
| Flux 3 Max | About 12 seconds | Tied for 1st (Elo 1,139, preliminary) | 20 seconds | No |
| FLUX 3 | About 126 seconds | Elo 1,089 | 20 seconds | No |
| Seedance 2.5 | About 278 seconds | Same reported Elo as Flux 3 Max | 30 seconds | Yes |
Sources: fal launch announcement and Artificial Analysis Video Arena, October 2026.
Creators who tried it in launch week kept coming back to the same four things.
A 10-second clip in about 12 seconds on fal means you can try five camera moves while other models finish one.
Cinematic lighting, skin detail, and small facial expressions hold up in close-ups and portrait shots.
Ambience, effects, and short lines of dialogue are generated with the picture. Switch audio off if you plan to add a voice-over.
Set the first and last frame and Flux 3 Max fills the motion between them, which keeps products and characters on model.
Six ways creators and marketers used it in launch week.
Selfie-style product clips and dozens of ad hooks to test, each rendered in seconds.
Use First & Last Frame on a packshot, turn sound off, and add your own voice-over.
Cinematic shots with sound for trailers, mood films, and short stories.
Block out every shot in Preview mode, then render the keepers with Flux 3 Max.
Painted, animated, and illustrated looks driven by the prompt.
Generate a move, extract the motion, and retarget it to a 3D character.
Launch-week posts on X from fal, Black Forest Labs, and the creator community, shown as the original posts.
fal announces FLUX 3 Max.
@fal · View on XBlack Forest Labs introduces FLUX 3 Max.
@bfl_ai · View on XArtificial Analysis Video Arena results for image-to-video and text-to-video.
@ArtificialAnlys · View on XA flying-Roadster car-soccer short made in two hours.
@PJaccetturo · View on XA movie trailer, with three tips for AI short films.
@OdinLovis · View on XUGC-style product videos rendered in about 25 seconds.
@VadimStrizheus · View on XA skateboarding-style short.
@mattworkman · View on XStylized animation, then upscaled.
@xikhar · View on XGenerated video turned into 3D motion capture and retargeted to a character.
@BlendiByl · View on XSubscribe monthly or yearly for a fresh balance every billing period, or buy a credit pack once. Every clip shows its credit cost before you generate, Preview mode costs less, and failed generations are refunded.
Launch price until Nov 7: a 10-second Flux 3 Max clip is 200 credits instead of 800.
STARTER
Save $24/yr$9.90$7.90/month
Billed $94.80 yearly
16,800 credits per year
Up to 84 Flux 3 Max clips (10s, 720p)
GROWTH
Save $72/yr$29.90$23.90/month
Billed $286.80 yearly
52,800 credits per year
Up to 264 Flux 3 Max clips (10s, 720p)
PRO
Save $144/yr$59.90$47.90/month
Billed $574.80 yearly
105,600 credits per year
Up to 528 Flux 3 Max clips (10s, 720p)
Credits are used when you generate a video. The cost depends on the clip length and resolution, and is always shown before you confirm.
Flux 3 Max is a video model released by fal on October 19, 2026, post-trained with Black Forest Labs on FLUX 3. It makes 5 to 20 second clips with sound from an image, a first and last frame, or a text prompt, at 720p or 1080p.
No. Flux 3 Max generates video. Earlier FLUX releases are best known as image models, which is why the name causes confusion, but Flux 3 Max outputs MP4 clips with sound.
No. Flux 3 Max has no public weights and runs through an API only. Black Forest Labs publishes some open-weight FLUX image models, but not this video model.
fal reports about 12 seconds to render a 10-second clip. Here, upload and queueing add a little time, so most clips are ready within a minute; 20-second and 1080p clips take longer.
Preview mode renders your prompt with the FLUX 3 draft model for fewer credits, so you can check motion and framing first. Switch it off to render the full Flux 3 Max clip.
Flux 3 Max is FLUX 3 post-trained by fal for speed and quality. fal reports it about 11.5 times faster than FLUX 3 for the same clip, with a higher arena score (Elo 1,139 versus 1,089).
Yes. Videos made with paid credits include commercial use. Only upload images you own or have permission to use, and label clips as AI-generated where a platform asks you to.
Not yet. Flux 3 Max covers text to video, image to video, and first and last frame. Video editing and reference-to-video are not available for it today.
Pick a mode in the workbench above and get a clip with sound in about a minute.
Start creating