FLUX 3 Video API Access: What Is Available Right Now

FLUX 3 was announced by Black Forest Labs on July 23, 2026. The video model has an early-access API, but that access is limited to initial partners. There is no public AI Flux3 integration today.

That short answer matters because a new model name quickly appears on landing pages before the underlying endpoint is broadly available. Here is the status we can verify without blurring an announcement into a working product.

Current access status

As of July 24, 2026:

• FLUX 3 Video API access is limited to initial partners.
• fal.ai does not list a FLUX 3 endpoint.
• Replicate does not list a FLUX 3 endpoint.
• AI Flux3 has not connected a FLUX 3 endpoint.

If you select FLUX 3 Early Access on this site, you are joining a waitlist. You are not sending a generation request to FLUX 3.

What has been announced

FLUX 3 Video is designed to accept text, image, and video inputs and produce video up to 20 seconds long with native audio. Announced controls include video continuation, keyframe transitions, multilingual dialogue lip sync, typography, and multiple connected clips.

Those details describe the intended capability surface. They do not tell us the public API schema, request limits, queue behavior, price, or output consistency. None of those operational details is present in our verified baseline, so we will not guess them.

Why fal.ai and Replicate matter

AI Flux3 uses third-party inference access to make models available through one studio. A provider endpoint is therefore the practical gate between “announced” and “usable here.”

We checked fal.ai and Replicate on July 24 and found no FLUX 3 endpoint on either service. This status can change after publication. The date on this article is part of the claim: it tells you when the check was true.

How to get access through AI Flux3

Join the FLUX 3 Early Access waitlist from the Studio. The waitlist exists for one purpose: to notify interested creators when a real integration is available.

Our release gate is straightforward:

1. A usable third-party FLUX 3 Video API must become available.
2. AI Flux3 must connect and verify that endpoint.
3. The Studio must label FLUX 3 outputs as FLUX 3.
4. Only then will the waitlist state change into a generation option.

This prevents an output from another model from being presented as FLUX 3.

What you can prepare now

You do not need an endpoint to prepare a generation brief. Gather the inputs the announced interface is designed to accept:

• A clear text description of the shot.
• A reference image when visual continuity matters.
• An existing video when the goal is continuation.
• Start and end frames for a planned keyframe transition.
• Dialogue and sound directions for native audio.

These are preparation steps, not a promise about the eventual API fields. The public request format may differ when access opens.

What happens next

FLUX 3 Image is expected within weeks, while an open FLUX 3 Dev edition is planned for later in 2026. Neither timeline provides a date for broad FLUX 3 Video API access.

Until an endpoint is live, the honest status remains: early partners only, no fal.ai endpoint, no Replicate endpoint, and no FLUX 3 generation inside AI Flux3.

FLUX is a trademark of Black Forest Labs Inc. AI Flux3 is an independent project and is not affiliated with Black Forest Labs.