What it is
FLUX 3 is Black Forest Labs’ multimodal model for generating video (optionally with native audio) plus images and action-related capabilities, available via dashboard/API with pay-as-you-go per-second video pricing.
Gabriel’s notes
FLUX 3 is Black Forest Labs’ multimodal foundation model aimed at generating and understanding video, audio, images, and actions within one unified architecture. As of July 23, 2026, Black Forest Labs describes FLUX 3 as available in Early Access, with FLUX 3 Video supporting text-to-video, image-to-video, and video-to-video flows (including keyframe-controlled transitions) and up to 20-second clips per generation with optional native audio.
Quick take: I haven’t personally tested FLUX 3 yet. But if it holds the line (or leaps the line) the way earlier FLUX releases did, it’s probably going to be a serious entry in the “make pixels move” arms race—especially for video with native audio.
I saved this under Video & audio because it’s squarely about short-form video generation (with optional built-in audio) and the operational reality of getting that into a production-ish workflow.
Also: if you’ve tried FLUX 3 and compared it to Google’s Veo / Gemini video stack (or anything else), send me a message. I genuinely want to hear how it behaves under real prompts, real constraints, and real deadlines—not just “look at my cinematic astronaut.”
Good fit if you want to:
- Prototype short video clips fast with a draft mode before paying for higher quality.
- Generate video with optional native audio (including dialogue and ambience) without stitching audio in later.
- Do image-to-video or video-to-video iterations where references actually matter (style, subject, continuity).
- Use an API (or a dashboard) rather than a purely consumer “one button, one mystery” UI.
- Track per-second unit economics instead of playing subscription-credit roulette.
Pricing snapshot (auto-enriched):
FLUX 3 Video is listed as pay-as-you-go with no subscriptions or seat fees. Current published rates on the FLUX 3 model page are per-second: text/image-to-video costs $0.06/sec (Draft HD), $0.17/sec (standard HD), or $0.29/sec (standard FHD); video-to-video costs $0.12/sec (Draft HD), $0.41/sec (standard HD), or $0.53/sec (standard FHD). Black Forest Labs also states that optional native audio is included at no extra charge.
Work-use / compliance snapshot (auto-enriched):
Black Forest Labs’ Terms of Service state that you grant them a broad license to access/use/modify Inputs and Outputs to provide, develop, train, and improve their services, with an opt-out mechanism by contacting their legal email (opt-out applies prospectively and has carve-outs for content already used, retained for safety/security/legal compliance, or submitted as feedback). If you’re self-hosting under a commercial weights license, their self-hosted commercial license terms include requirements around content filtering (or review before distribution), and they describe usage reporting/usage tracking expectations; they also prohibit distributing licensed models/derivatives to third parties via an API endpoint unless you contact sales for permission.
Alternatives (auto-enriched):
- Google Veo (via Gemini API): If you’re already in Google’s stack, Veo is a direct comparison point; Google publishes Veo pricing in their Gemini API pricing docs and posts periodic updates on Veo pricing/configurations via Google Developers Blog.
- Runway Gen-4: Runway is still the “creative suite + model” combo many teams use for iteration and finishing; they publish subscription pricing and also document API pricing/credit mechanics in their developer docs.
Before you adopt it:
- Run a tiny, boring benchmark: the same prompt, same reference assets, same duration, three times—then check consistency and failure modes, not just the best-of run.
- Decide upfront how you’ll handle IP/risk: prompts, reference images, and outputs can create rights and brand issues even when the tool is “technically allowed.”
- If you’re building anything client-facing, read the terms and plan your opt-out / retention posture early (don’t wait until after you’ve shipped).
Sources
- https://bfl.ai/models/flux-3
- https://bfl.ai/blog/flux-3
- https://bfl.ai/legal/terms-of-service
- https://bfl.ai/legal/self-hosted-commercial-license-terms
- https://ai.google.dev/gemini-api/docs/pricing