fal @fal
fal's launch post. Video of a polka-dot clown gunslinger strutting down a sunlit Western ghost-town street, 27 s montage.
Black Forest Labs fal Research
FLUX 3 Max is the video model from fal Research and Black Forest Labs. FLUX 3 Max video runs 5 to 20 seconds with sound, from text, a first frame or keyframes. Below: FLUX 3 Max image to video, the FLUX 3 Max price and FLUX 3 Max vs Seedance 2.5.
Tied #1 in image-to-video quality. Artificial Analysis timed about 12 seconds per 10 second clip. A 5 second clip is 530 credits at 720p. Launch price until Nov 7. See the charts or try it free.
Made with FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
Real clips. They play as you scroll; tap one for sound, or press Remix this to load its prompt in the try box.
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 Max
FLUX 3 MaxIn the image-to-video arena, FLUX 3 Max and Seedance 2.5 are statistically tied for first. The bars start at 1000 Elo.
Source: Artificial Analysis Video Arena, image-to-video Elo, as shown in fal's FLUX 3 Max launch chart, 19 October 2026. Elo is a score from head-to-head votes: a higher number won more of them. A 5 point gap is inside the noise.
Average time to make a 10 second video, as Artificial Analysis timed it for fal. Your own wait changes with clip length and load. H3 Max is faster; FLUX 3 Max is second on this chart, and far ahead of the rest.
Source: fal's FLUX 3 Max launch chart: average time for a 10 second 720p/768p 16:9 video. Your time includes the queue and depends on load.
Posts from people using FLUX 3 Max. They are not our clips. Nothing loads from X until you press Show the post.
fal @fal
fal's launch post. Video of a polka-dot clown gunslinger strutting down a sunlit Western ghost-town street, 27 s montage.
Black Forest Labs @bfl_ai
BFL launch post, square video: a mustachioed cowboy galloping a black horse through the desert at golden hour, 15 s.
PJ Ace @PJaccetturo
AI filmmaker PJ Ace: 48 s flying-car stadium sports spot, made in two hours with FLUX 3 Max.
Amira Zairi @azed_ai
Creator post with 18 s sample footage: an astronaut in a Western ghost town, close-up helmet portrait.
Blendi @BlendiByl
Open-source Prompt2Rig demo, 36 s: FLUX 3 Max video of a dancer is turned into 3D motion on a mannequin character. A developer-tooling use case.
Umesh @umesh_ai
Creator test with the prompt in the thread: moody rain-soaked blue-hour field, a woman in a long dark dress walking away.
NOBU @nbykos
Painterly anime-style schoolgirl on stairs in a flat teal and yellow palette, 12 s. Shows a stylised 2D look.
FLUX 3 Max is billed per second of video. These are the credits Ziztu charges, from the same quote the Generate button uses.
Launch price until Nov 7. fal sells FLUX 3 Max at 75% off until Nov 7 2026, and Ziztu passes it on. From Nov 7 a 5 second 720p clip is 2,110 credits. Reel Factory Max gives 50,000 credits a month: about 94 five-second 720p clips at the launch price, about 23 from Nov 7. See plans
| Length | 720p | 1080p |
|---|---|---|
| 5 seconds | 530 creditsfrom Nov 7: 2,110 credits | 900 creditsfrom Nov 7: 3,590 credits |
| 10 seconds | 1,060 creditsfrom Nov 7: 4,210 credits | 1,800 creditsfrom Nov 7: 7,170 credits |
| 15 seconds | 1,580 creditsfrom Nov 7: 6,310 credits | 2,690 creditsfrom Nov 7: 10,760 credits |
| 20 seconds | 2,110 creditsfrom Nov 7: 8,410 credits | 3,590 creditsfrom Nov 7: 14,340 credits |
A failed clip is refunded. The price is shown before every run.
FLUX 3 Max has three ways to start. All three make 5 to 20 seconds with native sound, at 720p or 1080p.
Write the shot and the sound. The model makes the picture. Shape: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 or 2:1.
In Ziztu: yes
Your picture is the first frame, and you can add a last frame to choose where the shot ends. Best when a character or a product must look the way you drew it.
In Ziztu: yes, first and last frame
Up to 10 pictures, each pinned to a frame number of the 24 frames per second clip. The model moves between them.
In Ziztu: through first and last frame; 10 keyframes on fal's API
The exact words, length and size of each clip above. Press Remix this to open the try box with the prompt and the first frame in place.
Two samurai clash in a bamboo forest (anime)
The two samurai clash again and again, blades ringing and sparks flying, then spring apart and slash back in, bamboo leaves bursting and spiralling around them, the camera pushes in and whips round the fight. 2D anime motion, smooth cel animation, strong impact frames. Sound: ringing sword clashes, rustling bamboo and leaves, a deep taiko drum hit. No text, captions, subtitles, logos or watermarks anywhere in the picture.
FLUX 3 Max · image to video · 5 s · 720p · 16:9. First frame made with GPT Image 2.5, then animated.

Perfume bottle on wet black stone
A wave of water splashes across the wet black stone around the perfume bottle in slow motion, droplets arcing and catching the studio light, ripples running through the thin water film, the camera pushes in slowly while a bright light sweeps across the faceted glass. Luxury product advertisement, smooth slow motion. Sound: a soft water splash and tinkling drops, a low shimmering synth pad. No text, captions, subtitles, logos or watermarks anywhere in the picture.
FLUX 3 Max · image to video · 5 s · 720p · 1:1. First frame made with GPT Image 2.5, then animated.

Hummingbird at a red flower, macro, 1080p
The hummingbird hovers at the red flower and dips its beak into the bloom, wings beating in a blur, it darts back a little and hovers again, tiny petals trembling, shallow macro focus, slow gentle camera drift. Photoreal nature documentary macro. Sound: the fast buzzing hum of wings, soft birdsong and a light breeze in the garden. No text, captions, subtitles, logos or watermarks anywhere in the picture.
FLUX 3 Max · image to video · 5 s · 1080p · 9:16. First frame made with GPT Image 2.5, then animated.
Interrogation room dialogue with native audio
Medium two-shot in a bare interrogation room, a metal table between two people, a single hanging lamp, cold concrete wall, a dark one-way mirror. On the left a nervous young witness in a grey hoodie, fidgeting with his hands; on the right a calm detective in a dark suit leaning forward. Static camera, shallow depth of field, moody cinematic lighting, photoreal. Witness (nervous, quiet, looking down): “I didn't see his face.” Detective (calm, low, leaning in): “Then tell me what you heard.” Sound: faint room tone and the buzz of the lamp. No text, captions, subtitles, logos or watermarks anywhere in the picture.
Dialogue with native audio. FLUX 3 Max · text to video · 7 s · 720p · 16:9. Text only.

Rainy night in a Tokyo ramen shop
A cosy rainy night inside a tiny Tokyo ramen shop, warm paper lantern light, steam rising from bowls on a worn wooden counter. A chef in a white headband lifts a ladle, tastes the broth with a loud slurp, closes his eyes and nods in approval. Rain runs down the window behind him with blurred neon glowing outside. Medium shot, slow gentle push-in, photoreal. Sound: soft rain on glass, bubbling broth, one clear slurp, low murmur of the shop, a faint warm radio tune. No text, captions, subtitles, logos or watermarks anywhere in the picture.
FLUX 3 Max · text to video · 5 s · 720p · 9:16. Text only.
Rider galloping along a misty Scottish ridge
Cinematic wide shot at dawn: a lone rider in a dark hooded cloak on a galloping chestnut horse racing along a misty Scottish highland ridge, heather and grey rocks, low golden sun breaking through drifting mist, the camera tracks alongside the horse at speed with a slight handheld sway, hooves throwing up clods of earth, mane and cloak streaming in the wind. Photoreal, anamorphic, epic. Sound: pounding hooves, rushing wind, a distant swelling orchestral string score. No text, captions, subtitles, logos or watermarks anywhere in the picture.
FLUX 3 Max · text to video · 5 s · 720p · 16:9. Text only.

Door opens on golden light (keyframes)
The girl stands still in the dim hallway looking at the closed door. The old wooden door slowly swings open and warm golden light pours through the doorway, filling the hall, glowing dust drifting in the beams, the girl lifts a hand towards the light, the camera pushes in slowly. Sound: a long wooden creak, a soft swelling warm choir chord. No text, captions, subtitles, logos or watermarks anywhere in the picture.
FLUX 3 Max · keyframes to video · 5 s · 720p · 9:16. Started from keyframes.
Drone rising over a neon cyberpunk city in rain
Cinematic drone shot rising slowly from a rain-soaked neon street, between glowing pink, cyan and violet signs (blurred, with no readable letters), up past flying cars and wet skyscraper walls to reveal a vast cyberpunk megacity skyline stretching to the horizon under heavy rain and low clouds, rain streaking through the light, glossy reflections below. Photoreal, moody, anamorphic. Sound: heavy rain, distant traffic hum, a deep pulsing synth drone. No text, captions, subtitles, logos or watermarks anywhere in the picture.
FLUX 3 Max · text to video · 10 s · 720p · 16:9. Text only.

Seaplane
The red seaplane skims fast across the lagoon straight toward the camera, spray fanning out behind the floats, propeller blur, the camera tracks low and tilts up slightly as it roars past, sun flare glittering on the water. Sound: loud radial engine roar and water spray.
FLUX 3 Max · image to video · 5 s · 720p · 9:16. First frame made with GPT Image 2.5, then animated.

Rooftop dancer
The dancer finishes a fast spin and strikes a pose, water splashing off the rooftop with each step, rain streaking through the neon light, the camera orbits slowly around the dancer. Sound: rain splashes, shoes slapping the wet floor, soft distant city hum, a faint beat.
FLUX 3 Max · image to video · 5 s · 720p · 9:16. First frame made with GPT Image 2.5, then animated.

Whale sailboat
The whale crashes back into the sea beside the sailboat, a huge wave of spray rushes over the deck, the boat rocks, the crowd on deck screams and cheers, the camera shakes slightly. Sound: a massive splash, the crowd screaming and laughing, waves and wind.
FLUX 3 Max · image to video · 5 s · 720p · 9:16. First frame made with GPT Image 2.5, then animated.

Anime rooftop jump
The girl sprints and leaps across the gap between the rooftops, the red scarf whipping behind her, the camera follows and tilts up as she flies over the drop, clouds glow in the sunset. Anime motion, smooth. Sound: running footsteps, wind rush, a swelling cinematic score.
FLUX 3 Max · image to video · 5 s · 720p · 9:16. First frame made with GPT Image 2.5, then animated.

Chef flambe
The chef tosses the pan, a huge fireball whooshes up above it and lights the kitchen orange, flames roll and die down, the camera pushes in a little, steam rising. Sound: loud pan sizzle, the whoosh of the fire, kitchen chatter and clattering.
FLUX 3 Max · image to video · 5 s · 720p · 9:16. First frame made with GPT Image 2.5, then animated.

Dog convertible
The golden retriever drives the convertible along the coast road, ears flapping in the wind, head turning to the camera with a happy pant, the sea sparkling behind, the camera moves alongside. Sound: wind, the car engine, ocean waves, a cheerful pant.
FLUX 3 Max · image to video · 5 s · 720p · 9:16. First frame made with GPT Image 2.5, then animated.
Bull rider boots ad (image to video, 1980s commercial look, quiet dialogue and narrator)
Continuous single shot, very slow push-in toward the rider’s boots. The young bull rider sits still on the bucket, head bowed, breathing slowly. He lifts his head slightly and looks down at his worn leather boots. A faint, tired smile. The arena behind him stays soft and blurred, the crowd quiet and still. Warm golden backlight, gentle lens flare, 1980s commercial look, 35mm film grain. Rider (quietly): “Lost my hat. Lost my pride. Kept my boots.” Narrator: “Ironside Boots. They stay on when you don’t.”
FLUX 3 Max · image to video · 10 s · 720p · 16:9. First frame made with an image model, then animated. Example by fal.
Western story across 4 keyframes: mesa, gallop, monsoon storm, campfire under the Milky Way
The lone cowboy waits on the mesa at golden hour, no move. then bursts into a full gallop across the red desert floor in a plume of dust, races along a dusty trail as a monsoon storm builds behind the buttes with rain curtains in the distance, and finally rests by a crackling campfire under the Milky Way. Epic orchestral western score, distant thunder, hooves and wind.
FLUX 3 Max · keyframes to video · 10 s · 720p · 16:9. Started from keyframes. Example by fal.
Credits for the same 5 second clip on Ziztu, from the live quote. Neither model is better at everything: run the same prompt on both and compare.
| Model | Length | Sound | Starts from | Credits per 5 s clip | Image-to-video Elo |
|---|---|---|---|---|---|
| FLUX 3 Max | 5 to 20 s | Yes | text, image | 530 credits720p | 1139 |
| Seedance 2.5 | 4 to 30 s | Yes | text, image, references | 2,930 credits720p | 1139 |
| Kling 3.0 Standard | 3 to 15 s | Yes | text, image | 1,560 credits720p | not in the chart |
| Kling 3.0 Pro | 3 to 15 s | Yes | text, image | 2,080 credits1080p | not in the chart |
Elo: Artificial Analysis Video Arena, image-to-video Elo, as shown in fal's FLUX 3 Max launch chart. Kling 3.0 is not in that chart, so there is no shared score. FLUX 3 Max is at its launch price until Nov 7. Seedance 2.5 page Kling 3.0 page
Copy a prompt, or press Try it to open the generator with the model and the prompt in place.
A red panda walks along a mossy log in a sunlit forest, pauses to sniff the air, then climbs over a fallen branch. Soft morning light through the canopy. Birdsong and rustling leaves.
Makes: A calm nature shot with ambient sound, from text alone.
A woman in a yellow raincoat crosses a night market after the rain. The camera follows from behind, then circles to her face as she tastes a skewer and smiles. Neon signs in the puddles. Soft chatter and sizzling.
Makes: A follow shot that ends on a close-up, with sound.
An old fisherman pulls a net onto a wooden boat at dawn. Low sun, mist on the water, slow push-in on his hands. Gulls and the creak of the boat.
Makes: A quiet 10 second shot with a slow push-in.
A courier on a bicycle weaves through a busy street market and shouts "Coming through!" as a crate of oranges tips over. Tracking shot from the side, bells and shouts.
Makes: A fast action beat with a spoken line.
A chef flips a pancake in a bright kitchen. Slow motion, the pancake rises, turns and lands in the pan. Static camera, a sizzle and a cheerful hum.
Makes: A slow motion action beat with food sounds.
Storm clouds roll over a lighthouse on a cliff, the sea crashing below. The light changes from gray to gold as the clouds part. Wind, waves and a distant gull.
Makes: An atmosphere change inside one shot.
FLUX 3 Max is in the AI Video Generator, next to the other video models, with the price shown before every run. One clip is one shot. For an episode, Ziztu Studio plans the shots, sends character sheets with every clip and lets you use FLUX 3 Max for the shots that suit it and another model for the rest, all on one credit balance. Start with the AI Micro Drama Generator or the AI Anime Video Generator.
Run the same prompt on a second model to compare the motion, the framing and the price.
FLUX 3 Max is a video model from fal Research and Black Forest Labs. It makes clips of 5 to 20 seconds with native sound from text, a first frame, a first and last frame, or up to 10 keyframes, at 720p or 1080p.
A new Ziztu account gets your first 5 second video free, no card, with a small watermark. More FLUX 3 Max video is paid. A 5 second clip at 720p costs 530 credits until Nov 7 2026 (fal's launch price) and 2,110 credits from Nov 7. Reel Factory Max gives 50,000 credits a month, about 94 of those clips at the launch price and about 23 after (see plans at /pricing). Black Forest Labs and fal do not publish a free allowance for it that we could source, so check their pages before you count on one.
On Ziztu a 5 second FLUX 3 Max clip at 720p costs 530 credits until Nov 7 2026 (fal's launch price) and 2,110 credits from Nov 7. At 1080p it costs 900 credits until Nov 7 and 3,590 credits after, and a 10 second 720p clip costs 1,060 credits until Nov 7 and 4,210 credits after. Price scales with length: it is billed per second of video. The Generate button shows the exact price before you run. See plans at /pricing.
Both sit at 1139 Elo in the image-to-video arena in fal's launch chart, a statistical tie for first. FLUX 3 Max makes 5 to 20 second clips and was timed by Artificial Analysis at about 12.45 seconds on average for a 10 second clip in that chart; Seedance 2.5 makes 4 to 30 seconds, takes up to 9 reference pictures in Ziztu and averaged 277.8 seconds. On Ziztu a 5 second 720p clip is 530 credits on FLUX 3 Max until Nov 7 (2,110 credits after) and 2,930 credits on Seedance 2.5. Pick FLUX 3 Max for speed and a quick check of an idea, Seedance 2.5 for clips longer than 20 seconds or when you need reference pictures.
They are different tools. fal's launch chart does not include Kling 3.0, so there is no shared score to quote. FLUX 3 Max makes 5 to 20 second clips at 720p or 1080p from text, a first frame, a last frame or keyframes. Kling 3.0 makes 3 to 15 second clips and keeps a cast steady with character elements. On Ziztu the lowest Kling 3.0 clip (5 seconds) costs 1,560 credits against 530 credits for FLUX 3 Max at 720p until Nov 7 (2,110 credits after). The fair test is to run the same prompt on both in the generator and compare.
From 5 to 20 seconds in one clip, in whole seconds. A 20 second clip at 720p costs 2,110 credits on Ziztu until Nov 7 2026 and 8,410 credits after. For a longer film, cut several clips together; Ziztu Studio plans the shots and keeps the cast the same across them.
Yes. FLUX 3 Max makes native audio with the picture: sound effects, ambience and speech, and the sound switch is on by default. Write the sound you want in the prompt, such as "loud engine roar and spray" or a line of dialogue in quotes.
Yes. It makes 720p and 1080p. 720p is the standard size and 1080p costs more: 900 credits for 5 seconds against 530 credits at 720p on Ziztu until Nov 7 (3,590 credits against 2,110 credits after). Draft at 720p, then rerun the prompt you like at 1080p.
1. Open the try box at the top of this page, or the AI Video Generator, and pick FLUX 3 Max. 2. Add a first frame (a picture of your scene, character or product) and, if you want, a last frame. 3. Write what happens, how the camera moves and what we hear. 4. Choose 5 to 20 seconds and 720p or 1080p, check the price, and press Generate.
Ziztu is the app, not an API. Developers can call FLUX 3 Max through fal, which hosts it: the endpoints are blackforestlabs/flux-3-max/text-to-video, blackforestlabs/flux-3-max/image-to-video and blackforestlabs/flux-3-max/keyframes-to-video, read on 20 October 2026. Check fal for its own price and terms.
Try FLUX 3 Max and see the price first.
400 free credits and your first 5 second video free. No card. Free clips carry a small watermark. More video starts on a plan: Reel Factory Max gives 50,000 credits a month.