Génération de vidéos
Disponible maintenant

FLUX 3 Max

par fal.ai

fal's post‑trained FLUX 3.
Tuned to follow the script, built for speed.

FLUX 3 Max

Fonctionnalités clés

Un kitesurfeur s'envole d'une vagueEssayer maintenant

Render times are fal's published benchmark, plus two timed runs of our own on Morphic. Every clip on this page was made with FLUX 3 Max on Morphic.

Spécifications techniques

~12 s

fal's average for a 10-second 720p clip

5–20 s

Any whole second, set per clip

720p / 1080p

720p by default, 1080p when you need it

2–10

First and last frame, or timed stills

Cas d'utilisation

Ad scenes from a script

A courier skids up, hands over the bag, and both laugh, in that order. Write the beats once and the clip plays them through.

Physical detail up close

Ice cracks as it settles and tea splashes over the rim, with each sound landing on the moment you can see it happen.

Spoken lines on camera

Put the line in quotes and it is spoken word for word, lip-synced, over the wind and the gulls of the dock around it.

Long single takes

A 15-second drone follow holds the car, the road, and the reveal of the lighthouse in one move, without a cut or a stitch.

Keyframed transformations

Three stills of clay, cylinder, and bowl become one continuous throw on the wheel, with each stage landing where it was pinned.

Stills brought to life

Hand it a photo and it moves inside that frame: the carousel turns, the bulbs glow, and the sea and gulls fill in the sound.

Exemples de prompts

Script in beats

A florist rolls up her shutter, sets out tulips, then grins at the camera

Edit prompt

Quoted line

A watchmaker lifts his loupe and says: Every watch tells you how it was treated

Edit prompt

Sound-led shot

A table tennis rally in an empty gym, every hit and bounce crisp

Edit prompt

Ultrawide take

21:9, a lone hiker on a snowy ridge under a vast evening sky, wind

Edit prompt

Overview

FLUX 3 Max is a video model from fal Research, built on Black Forest Labs' FLUX 3 and post-trained by fal, in partnership with Black Forest Labs, for closer prompt adherence, better-looking output, and speed. It launched on October 19, 2026. A clip runs 5 to 20 seconds at 720p or 1080p, with sound generated alongside the picture.

What FLUX 3 Max does differently

The headline is time. fal's model page puts a 10-second clip at about 12 seconds of rendering, where video at this quality usually takes minutes, so a draft becomes something you watch, adjust, and run again in one sitting. The post-training shows too: actions written in a row play in order, and on Artificial Analysis' image-to-video ranking it sits level with the top models. Max is fal's tuned edition of FLUX 3, not a bigger tier from Black Forest Labs.

FLUX 3 Max on Morphic

FLUX 3 Max sits in the video model picker beside Seedance, Kling, Veo, and MiniMax H3, for text to video, image to video, and frames to video with up to ten keyframes. Morphic runs it through fal, the team that post-trained it, at fal's high reasoning setting, the one that follows a prompt most closely. Takes land on the Canvas, so you can run the same prompt through another model and compare them side by side. For how to write for it, see the FLUX 3 Max guide, and for ready-made prompts, the FLUX 3 Max prompt library.

Tarifs simples

Commencez gratuitement dès aujourd'hui, avec la possibilité de mettre à niveau ou d'annuler à tout moment.

Basic

$19/ mois
facturé comme $0 par an

2400 mensuel crédits

1 utilisateur uniquement

Tous les modèles

Workflows

Standard

$24/ mois
facturé comme $0 par an

3625 mensuel crédits

1 utilisateur uniquement

Tous les modèles

Workflows

Pro

$45/ mois
facturé comme $0 par an

6350 partagé mensuel crédits

1 utilisateur

+ jusqu'à 4 plus à un coût supplémentaire

Tous les modèles

Workflows

Max

$170/ mois
facturé comme $0 par an

24650 partagé mensuel crédits

1 utilisateur

+ jusqu'à 9 plus à un coût supplémentaire

Tous les modèles

Workflows

Enterprise

Pour des limites plus élevées

Personnalisé

conditions de tarification et de facturation

Crédits à haut volume
Limites de sièges personnalisées
Tous les modèles
Workflows
Pricing Gradient

Free

Pour expérimenter

$0

gratuit pour toujours

Jusqu'à 20 crédits
1 utilisateur uniquement
Modèles limités
Workflows
Comparer les détails des plans

Questions fréquentes

What is FLUX 3 Max?
FLUX 3 Max is a video model from fal Research, built on Black Forest Labs' FLUX 3 and post-trained by fal for closer prompt adherence, better-looking output, and much faster generation. It launched on October 19, 2026, and makes 5 to 20 second clips with native audio from text, an image, or keyframes.
Is FLUX 3 Max made by Black Forest Labs?
Black Forest Labs built FLUX 3, the foundation model. fal Research did the post-training, reinforcement learning, and inference work that turned it into FLUX 3 Max, in partnership with Black Forest Labs, and fal serves it. So Max is fal's tuned edition of FLUX 3, not a larger FLUX 3 tier released by Black Forest Labs.
How is FLUX 3 Max different from FLUX 3?
It is tuned to follow the prompt more closely and to look better, and it is far faster: fal reports more than 11 times faster inference and a 50-point Elo gain over FLUX 3. It covers text, image, and keyframes to video. Continuing or editing an existing clip, and the 2K and 4K output of Black Forest Labs' own API, stay with FLUX 3.
How fast is FLUX 3 Max?
fal's benchmark puts a 10-second 720p clip at about 12 seconds of rendering, which it says is about 22 times faster than Seedance 2.5 at the same quality. On Morphic, two 10-second clips with sound each came back in under 30 seconds, queue included.
How good is FLUX 3 Max compared with other video models?
On the Artificial Analysis image-to-video leaderboard at launch it scored about 1,138 Elo, level with Seedance 2.5 and two points behind MiniMax H3, with overlapping confidence intervals. That makes it a peer of the top models on quality rather than a clear winner, and the results are preliminary and cover image to video only.
Does FLUX 3 Max generate audio?
Yes. Sound is generated with the picture in the same pass: effects, ambience, and dialogue written in quotes, lip-synced to the speaker. In our tests on Morphic, all six quoted lines we prompted came back word for word. You can switch audio off for a silent clip.
What lengths, resolutions, and aspect ratios does FLUX 3 Max support?
Any whole-second length from 5 to 20 seconds, at 720p by default or 1080p. Text to video takes seven aspect ratios: 21:9, 2:1, 16:9, 4:3, 1:1, 3:4, and 9:16. Image to video and keyframes follow the shape of the stills you give it.
How do I use FLUX 3 Max on Morphic?
Pick FLUX 3 Max in the video model picker, or name it in a Copilot prompt. Write the shot as a sequence of actions with the camera and the sound, or add a still to animate, a first and last frame, or up to ten keyframes. FLUX 3 Max is available on Morphic's paid plans.

Autres modèles

OmniHuman 2.0

ByteDance

Le modèle vidéo de ByteDance piloté par l’audio. Une photo, une piste, cinq minutes en un seul plan.

Mirelo SFX 2.0

Mirelo AI

Le modèle vidéo vers son de Mirelo. Un nouveau son pour votre vidéo, les mêmes dialogues.

Fish Audio Drama 3

Fish Audio

Le modèle vocal le plus expressif de Fish Audio. Vous dirigez le jeu, mot après mot.

Vidu Q4

Shengshu Technology

Dernier modèle vidéo de Shengshu. Jusqu’à 12 images de référence et 3 voix tenues sur tout le clip, audio natif, du 540p à la 4K.