Standard preview
$1.02
5s · 720p
FLUX VIDEO WITH TEXT AND FRAME CONTROL
Turn text or images into 5–20 second videos with optional audio and first-and-last-frame guidance.
Direct a cinematic scene, a product film or a talking portrait. Start with your prompt or provide the composition you want to animate. Add an ending image when the final frame matters, then choose your duration, format and audio before reviewing the price.
Start with text or an image
Describe a new scene, animate one image, connect first and last images, or place keyframes throughout a sequence.
Shape your scene’s rhythm
Choose any whole-second duration from 5 to 20 seconds. Leave room for action, dialogue and the pauses that give the scene its rhythm.
Direct the sound
Generate dialogue, ambience or music with the scene, or turn audio off. Put the words and sounds you want directly in your prompt.
Choose the delivery format
Use landscape, portrait, square or wider framing. 720p is the default; 1080p uses upscaling. High reasoning is the default, with low also available.
Compare five seconds at 720p and 1080p, then a ten-second 720p scene.
$1.02
5s · 720p
$1.74
5s · 1080p
$2.04
10s · 720p
Generation example: 5 s · 720p · Text to video · Audio included — $1.02
The per-second figure is the effective rate for the five-second example. Each clip total is rounded to the cent. Review the current quote before generating.
Describe the action, the progression between shots and the rhythm of the reveal. Adapt framing, light and sound to the genre; exact cuts are not guaranteed.
Begin with a clear image and describe a small change of expression or pose while preserving the subject and composition.
Use two compatible images of the same subject or space. Explain the movement or lighting change that connects them.
Copy the exact English prompts behind four original films. Shape the editing, acting and sound around the rhythm of each genre.
Name the subject, visible action, camera movement and sound. No reference image is required.
Provide the opening image, then direct movement without redesigning the face, product or scene.
Provide both images and describe one continuous path between their compositions.
Arrange 1–10 images at selected positions in the clip to guide how the scene changes over time.
A woman reaches cover in a concrete parking garage, then answers over her radio.
A grounded action-thriller scene in an empty concrete parking garage at night. An adult woman with short dark hair, wearing a charcoal jacket and dark trousers, sprints from left to right toward a thick square concrete pillar. A burst of gunfire from offscreen behind her chips the wall she has just passed. She reaches the pillar, turns around its near corner, and presses her back against the sheltered side. The firing stops. Dust settles. Hold close on her face as she catches her breath and listens. She lifts the radio clipped to her jacket and says quietly in English, "I'm still here." Cut to black. Start with a clear wide view of the route and cover; use urgent cuts following her run, then let the final pause last. Preserve the same woman, clothes, pillar and screen direction throughout. Hard fluorescent light, a red emergency light in the distance. Sound: running footsteps, offscreen gunfire, concrete impacts, breathing, radio static and her voice. No music or visible injuries.
Requested settings: text to video · 14 s · 16:9 · 720p · audio on.
A woman reaches cover in a concrete parking garage, then answers over her radio.
Best practices, common fixes, and important limitations to help you get the strongest results with FLUX 3 Max.
Compare prices, formats, and capabilities before choosing your model.
Generate FLUX 3 video from text, images, first/last frames, or extend an MP4.
Compare FLUX 3 Max vs FLUX 3Formats, settings and limits to check before creating.
Built-in safeguards and best practices for responsible creation with FLUX 3 Max.
FLUX 3 Max is a video model post-trained by fal in collaboration with Black Forest Labs, based on the FLUX 3 Video family. On MaxVideoAI, it creates videos from text, an opening image, first and last images, or a timed sequence of keyframes.
Yes. Choose FLUX 3 Max in the MaxVideoAI web workspace, write a prompt, add images if your mode requires them, and select your settings. Review the quote and generate from your account; you do not need your own fal API key.
Audio generation is optional and enabled by default. Describe the dialogue, ambience or music in the prompt, or turn audio off for a silent video.
Choose a whole-second duration from 5 to 20 seconds. The default output is 720p; the 1080p option uses upscaling. Landscape, portrait, square and wider aspect ratios are available.
Yes. Image-to-video starts from one image. First-and-last-frame generation uses two supplied images to guide the beginning and ending. Keep the subject, perspective and geometry compatible, and describe how the scene moves between them.
Max is fal’s post-trained variant of FLUX 3 Video. This MaxVideoAI integration supports text-to-video, image-to-video, first-and-last-frame generation and timed keyframes. Standard FLUX 3 also has a source-video extension workflow; extension is not included in the Max integration.
The price depends on the duration, resolution and selected settings. This page shows current scenario prices from MaxVideoAI. Your generation quote confirms the amount before you spend credits.
In fal’s published comparison, a 10-second 16:9 clip at 720p took 12.45 seconds to generate. This is a provider benchmark, not a measured MaxVideoAI delivery time. Your wait also depends on settings, queueing and processing.
Artificial Analysis snapshot, 20 October 2026: FLUX 3 Max (720p) has 1,138 Elo in image-to-video with audio and 1,159 in text-to-video with audio. The two boards have separate Elo scales and overlapping confidence intervals with leading models.
Editorial estimate, 20 October 2026: our eleven scores use selected public fal examples, documented controls and prices, with Artificial Analysis as overall preference context. We did not run a controlled MaxVideoAI test set. Confidence varies by criterion; lip sync and delivery reliability remain weakly evidenced. The 8.7/10 overall is an editorial estimate, not an Elo conversion.
The Prompt Lab shows the exact English prompts for four original films: action, romance, science fiction and a fantasy keyframe scene. Each prompt links to its own rendered example and reuse action. The keyframe demonstration also shows its three input images and positions.
Provide 1–10 images and place each one at a chosen position in the clip. Describe the action and camera movement between them. These are timed visual anchors for a sequence, rather than free-form reference images for a character.