FLUX.3

flux-3

FLUX.3

flux-3

Generate up to 20-second videos with native audio from a prompt, a start image, start/end frames, multiple keyframes, or a source clip. FLUX.3 chooses the matching workflow automatically from what you attach.

Added Aug 5, 2026

Approx. Price

$0.300 per video

Model Type

both

Settings

Generation controls available for this model.

Output Format

up to 1,080p
Auto
Square
Portrait
Landscape

Default Duration

5

16 duration options

Aspect Ratio

Select

Default

auto

Options (8)

Auto, Ultra-wide (21:9), Wide (2:1), Landscape (16:9) +4 more

Output frame shape; Auto chooses from the attached media or prompt.

Duration

Select

Default

5

Options (16)

5 seconds, 6 seconds, 7 seconds, 8 seconds +12 more

Clip length from 5 to 20 seconds.

Generate Audio

Toggle

Default

Yes

Generate synchronized audio and dialogue.

Render Quality

Select

Default

full

Options (2)

Full quality, Draft preview

Draft creates a lower-cost preview; Full enables the highest output quality.

Resolution

Select

Default

720p

Options (2)

720p, 1080p

1080p is available for full-quality generations.

Benchmarks

Human preference benchmarks sourced from LMArena.

Text to Video

#4 / 48

Arena Score

1493.8

Votes

1,290

Confidence Interval

1476.5 - 1511.1

Image to Video

#8 / 47

Arena Score

1448.6

Votes

20,601

Confidence Interval

1442.3 - 1454.9

Published 2026-09-04 · Matched as flux-3-video

LMArena Dataset

Examples

Loading examples…

Compare FLUX.3 with similar models from the same provider or model family.

FLUX.3 Edit Video

blackforestlabs/flux-3/edit-video

Edit existing footage with natural-language instructions. Change objects, weather, or the look of a scene while preserving motion, timing, and framing. Requires an MP4 under 15 seconds and 50 MB; output is 720p.

MiniMax H3 Max Multi Angle

minimax/h3-max/multi-angle/image-to-video

Animate a starting image with precise camera control: orbit around the subject, move closer, pull back, or rise above the scene. Choose a camera movement or supply custom keyframes. Supports 5–15 seconds at 480p, 768p, or refined 1080p.

LTX-2.5 Fast

lightricks/ltx-2.5/fast

Speed-optimized audiovisual generation from text, an image, or a 2-20 second audio clip. Creates synchronized video and audio in one pass, with output up to 4K and optional start/end-frame control.

LTX-2.5 Pro

lightricks/ltx-2.5/pro

High-fidelity audiovisual generation from text, an image, or a 2-20 second audio clip. Creates polished synchronized video and audio in one pass, with 720p/1080p output and optional start/end-frame control.

Pixelcut Video Background Remover

pixelcut/video-background-removal

Remove video backgrounds with frame-by-frame AI segmentation and temporally consistent edges. Supports transparent output, preset solid backgrounds, custom RGB backgrounds, and common video formats.

Bernini R Video

bernini-r-video

Bernini R video generation and editing. A single NanoGPT model id routes prompt-only requests to text-to-video, up to 5 image inputs to reference-to-video, video inputs to edit-video, and image plus video inputs to reference-edit-video.