FLUX 3 Max API - Video Generation with Native Audio

FLUX 3 Max is an optimized video generation model built on the FLUX 3 video foundation. It creates video from text, still images or positioned keyframes, with native audio generated alongside the visuals. Its strengths include combined visual and audio generation, first-and-last-frame guidance and multiple positioned keyframes. Through the PoYo API, it supports 5–20 second clips at 720p or 1080p with optional audio for image animation, product showcases and storyboarding.

Get an API key from your PoYo dashboard, add credits, and choose a generation mode based on your source material. Use flux-3-max/text-to-video for a written scene, flux-3-max/image-to-video for an opening image and optional ending image, or flux-3-max/keyframes-to-video for positioned images. Test your prompt, duration, resolution and sound setting in the Playground before using the same inputs in an API request. Save the returned task ID and query its status to retrieve the completed video, using only the media fields for the selected mode.

Text-to-video, image-to-video and keyframe generation share the same rates: $0.17 per second at 720p and $0.29 per second at 1080p. A 5-second clip therefore costs $0.85 or $1.45, and a 10-second clip costs $1.70 or $2.90, respectively. Choose a duration that gives your action enough time, select a resolution for your delivery needs, and review the estimate for your current configuration in the Playground before submitting.

Choose an integer duration from 5 to 20 seconds, with 5 seconds as the default, and either 720p or 1080p output, defaulting to 720p. The aspect_ratio options are auto, 21:9, 2:1, 16:9, 4:3, 1:1, 3:4 and 9:16 for widescreen, landscape, square and portrait compositions. With the default auto setting, text-to-video uses 16:9, while image and keyframe modes follow the first image. Match the format to your intended placement and allow enough time for the subject action and camera movement.

In image-to-video mode, provide 1–2 images through image_urls: the first sets the opening frame, and an optional second image guides the ending frame. Use the prompt to describe the action and camera movement between them. For example, start with a closed gift box, end with an open box, and describe the lid slowly opening. This guidance is useful for product reveals, scene changes and action transitions; prepare visually coherent subjects and settings to make the intended change clear.

Keyframe mode accepts 1–10 entries in keyframes, each containing image_url and an integer frame_index. Indices must be unique and fall between 0 and duration × 24, inclusive. For a 5-second clip, positions 0, 60 and 120 can guide the opening, midpoint and ending. Use the prompt to describe the actions, camera movement and environmental changes between these visual anchors. Before submitting, check that every position fits the selected duration and that the reference images present coherent subjects and settings.

Yes. FLUX 3 Max can generate native audio alongside the video, with sound set to true by default. Set it to false when you need silent footage for separate voiceover or editing. Your prompt can describe dialogue, delivery, ambience and action sounds, such as a character speaking while rain and footsteps are heard. Describe the intended sounds together with the corresponding scene actions to communicate your creative direction, then review the full clip for speech content, sound timing and visual performance before selecting it.

On PoYo, FLUX 3 Max provides text-to-video, image-to-video and keyframe generation, with first-frame and optional last-frame guidance included in image-to-video. It suits workflows that create new clips from written scenes, still images or storyboard plans. The FLUX 3 family also offers a separate first/last-frame endpoint and an extend-video endpoint for continuing existing footage. Both support native audio and 720p or 1080p output. Choose the FLUX 3 extension interface when your workflow needs video continuation; the current Max endpoints do not accept an existing video for extension.