Skip to content
English

FLUX 3 API

FLUX 3 is the next-generation model family from Black Forest Labs, announced on July 23, 2026. The first announced release is FLUX.3 Video, a video generation model designed around native audio, multimodal control, and longer structured sequences. FLUX.3 Image is planned for a later Early Access release.

The following capabilities come from the official FLUX.3 announcement and FLUX 3 product page:

AreaOfficially announced capability
DurationUp to 20 seconds of generated video
AudioNative audio, including multilingual dialogue
Text controlText-to-video generation
Image controlImage-to-video with a start frame or reference image
Video controlVideo-to-video generation
ContinuationVideo and audio continuation
Timeline controlKeyframe-to-video generation
Story structureMulti-shot chaining
AvailabilityFLUX.3 Video Early Access; public product page marked Coming Soon

The production API contract is not public yet. HiAPI will not guess or copy parameters from another FLUX model. The following details remain pending:

  • Production model IDs and available variants
  • Request field names, required inputs, enums, defaults, and limits
  • Resolution, duration, audio, and reference-media combinations exposed by the API
  • Pricing and billing dimensions
  • Final availability date

Until those details are published, there is intentionally no provisional request example on this page.

  • Native-audio short films — generate picture, ambience, effects, and multilingual dialogue as one coordinated result.
  • Reference-guided animation — start from a frame or reference image while preserving the intended subject and visual direction.
  • Video transformation and continuation — restyle existing footage or extend a video and its audio beyond the source clip.
  • Keyframed and multi-shot sequences — control important timeline states and chain multiple shots into a structured sequence.

HiAPI has reserved the model ID flux-3 as a Coming Soon entry. The public parameter schema, task mapping, and pricing will only be added after the production API is available and verified end to end.

At launch, this page will be updated with:

  • The exact /v1/tasks request schema
  • Copy-ready curl, Python, and Node.js examples
  • Input constraints and supported media formats
  • Live pricing and billing examples
  • Tested output and polling behavior

Not yet on HiAPI. Black Forest Labs lists the product as Coming Soon, and FLUX.3 Video is currently in Early Access.

The official announcement states up to 20 seconds. The exact duration options exposed by the production API have not been published yet.

Yes. Native audio and multilingual dialogue are confirmed capabilities. Exact audio controls and supported combinations will be documented after the API contract is published.

Does FLUX 3 support image-to-video and video-to-video?

Section titled “Does FLUX 3 support image-to-video and video-to-video?”

Yes. Black Forest Labs has announced image-to-video with start-frame or reference-image control, plus video-to-video generation.

No. This Coming Soon entry tracks FLUX.3 Video. Black Forest Labs says FLUX.3 Image Early Access is planned later, and it will be documented separately once its API surface is known.

Keep the generation workflow behind a model-specific adapter and use HiAPI’s unified async API for task lifecycle handling. Do not hard-code FLUX 3 input fields until the verified schema is published here.