FLUX 3 Max on VisionX.
A post-trained FLUX 3 Video, tuned to follow the prompt and to render sooner.
What FLUX 3 Max is on the roster
FLUX 3 Max is a post-trained version of FLUX 3 Video, tuned for stronger prompt adherence and better visual quality, and built to come back with lower latency. It renders one continuous take of five to twenty seconds, with the sound generated in the same pass as the picture.
It takes text, a first frame, or a first and a last frame, so a shot can open on a still you have approved and land on another. It renders at 720p and 1080p, and a reasoning setting lets you trade some prompt-following for speed while you are still finding the shot.
Closer to the prompt
Post-trained on top of FLUX 3 Video for stronger prompt adherence and better visual quality, with the same one-take, sound-on behaviour.
Five to twenty seconds, one take
Any whole second from five to twenty, rendered as one continuous shot rather than assembled from several.
Open on a frame, land on a frame
Give it a first frame, or a first and a last frame, and the shot starts on the image you approved and can end on another.
A faster draft when you want one
A reasoning setting trades some prompt-following for speed, so a draft can come back sooner before you commit to the keeper.
What FLUX 3 Max actually does
The post-trained FLUX 3 Video, and what it takes.
Closer to the prompt
Post-trained on top of FLUX 3 Video for stronger prompt adherence and better visual quality, and tuned for lower latency.
Post-trainedFive to twenty seconds, one take
Any whole second from five to twenty, as a single continuous shot long enough to carry a full beat.
Long takeA first frame, and a last
Open the shot on a still, or on one still and end it on another, and the model works out the motion between them.
FramesSound rendered with the picture
Audio is generated in the same pass and can be switched off per shot, with no difference in price either way.
Native audioChoose how hard it reads the prompt
High reasoning, the default, follows the prompt closely. Low reasoning comes back sooner, which suits a draft. The model expands the prompt itself on either setting.
Reasoning720p or 1080p, six ratios or auto
Pick a rung and it is quoted per second before you run it. Choose 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16, or leave aspect on auto: it is 16:9 for text and follows your first frame otherwise.
Resolution and aspectEvery variant, exactly as the studio runs it
These rows are read live from the engine registry and the pricing engine. What you see here is what the composer quotes.
Self-serve rates run as low as $0.085 per VX at scale. The composer quotes the exact VX for your shot, references included, before you run it.
How to write for FLUX 3 Max
A frame carries the composition. The prompt carries the motion and the sound.
Say what moves
With a first frame attached, describe the motion and the camera rather than describing the picture again.
Describe the sound
Audio renders with the picture, so ambience and effects belong in the prompt. Switch audio off when the cut will be scored.
Expect the prompt to be expanded
The model reads and expands your prompt before it renders, on either reasoning level. Write the beats you need and anchor identity and style with a frame instead of relying on exact wording.
Draft on low, finish on high
Use low reasoning to find the shot, then re-run the one you keep on high.
What teams route to FLUX 3 Max
Where FLUX 3 Max earns the pick.
A shot that has to start on a hero frame
Open on the approved product or character still and let the model carry it into motion.
Start and end boards
Pin the first and last boards of a beat and let the model animate between them.
Drafts at speed
Find the shot on low reasoning, then run the keeper on high.
Same Cast, same wallet, different look
Route a board line to FLUX 3 Max, then run the identical shot on another family and compare takes. Identity lives in your Cast, not in any one model, so switching engines never strands the campaign.