FLUX 3 review, pricing and verdict

FLUX 3 launched July 23, 2026 as the first model from Black Forest Labs, the 2024-founded lab behind FLUX.1 and FLUX.2, to jointly generate video, image, and audio from one Self-Flow architecture. It suits teams needing synchronized dialogue baked into the video itself, rather than dubbed on afterward.

FLUX 3 is Black Forest Labs' unified multimodal model, built on the Self-Flow architecture, generating twenty-second video clips at resolutions up to 4K with native multilingual audio, dialogue, and sound effects from text, images, or existing clips. The family also includes FLUX 3 Image and the open-weights FLUX 3 Action.

Provider: Black Forest Labs · Family: FLUX 3

More about Black Forest Labs on Hokai

Input modalities: text, image, video · Output: video, audio

About FLUX 3

FLUX 3 is Black Forest Labs' first multimodal foundation model, announced July 23, 2026 and built on a new architecture the company calls Self-Flow, which jointly trains on images, video, and audio inside one shared system rather than bolting separate generators together. Black Forest Labs is the German AI lab founded in 2024 by former Stability AI researchers behind the original Stable Diffusion and FLUX.1 image line, and FLUX 3 is their move from still images into a single model that also produces motion and sound. The design goal, per the company's own launch material, is a model that learns how objects hold together, how things move, and how events sound, instead of learning each modality in isolation.

FLUX 3 Video runs on the Black Forest Labs API, and the vendor's documentation still describes FLUX 3 as a preview model. A single request returns up to 20 seconds of 24 fps video with synchronized audio, from a text prompt, from keyframes (one image is a start frame, two pin the start and end, up to ten storyboard the clip, and each can be pinned to an exact second), or from an existing clip to continue. Output starts at HD (720p) and goes up to 4K, 3840 by 2176 pixels for 16:9, where the fhd, qhd and uhd classes are finished by a video upsampler; supported aspect ratios run from 21:9 to 9:21. The documented modes are text to video, image to video and video continuation, with multilingual speech and lip-sync, effects and ambience generated alongside the frames. A separate FLUX Video Edit [fast] endpoint applies a text instruction to a source clip and keeps its length. Draft mode returns a fast preview that costs about a third of a full render, and a draft can be re-rendered at full quality with the same shot and seed.

No independent third-party benchmark suite exists for FLUX 3 the way SWE-bench or MMLU exists for language models. Black Forest Labs published its own pre-release preference-test win rates against two named rivals ahead of launch, but those figures come from an early checkpoint and were not independently reproduced, so no benchmark figures are recorded here until a third party publishes one.

FLUX 3 Video is billed per second of output rather than per token. Rates rise with the resolution class (HD, Full HD, QHD, UHD) and run higher for video continuation than for text or image input, and drafts have their own lower rate. There is no subscription or seat fee: usage is metered against a prepaid balance, with the same price in the API and the Playground. The still-image products carry their own per-image rate cards, and the open-weights Klein image model is the cheapest of them; exact video rates are in the pricing table below.

Access today is the direct Black Forest Labs API (docs.bfl.ai) plus third-party inference platforms including fal.ai, and the vendor's product page names creative tools such as Picsart, Krea and HeyGen among its partners, and a Nous Research testimonial describes Hermes Agent generating a shot, checking it and continuing from it. Third-party rates were not re-checked for this page and can differ from the vendor's. The vendor's documentation does not list FLUX 3 on AWS Bedrock, Google Vertex or Azure; the company's prior FLUX.1 and FLUX.2 image lines are more broadly distributed across those clouds.

Beyond video, the FLUX 3 documentation lists FLUX 3 Action, a 7B world action model with open weights: given camera images, the current state and an instruction, it predicts a chunk of actions, and it can be run on a robot arm or fine-tuned on a robot or game. FLUX 3 Image Klein is also released with open weights. This is the clearest signal of what Self-Flow is for: a shared world model that outputs pixels, sound or motor commands depending on the head attached to it.

No public system card, red-team partner disclosure, or safety benchmark has been released for FLUX 3. Black Forest Labs has not published a HarmBench score, jailbreak-resistance rating, or content-restriction policy specific to FLUX 3, unlike frontier LLM vendors who publish these alongside launch. Buyers evaluating FLUX 3 for production should treat the safety posture as unverified rather than assume parity with the company's FLUX.1/FLUX.2 image-model documentation.

FLUX 3 is best suited today for teams building short-form video with synchronized dialogue or sound baked into the generation itself, rather than as a separate TTS or foley pass, and for teams that specifically need the video-continuation and keyframe-transition features. It is not yet a fit for teams that need a non-preview commitment, for teams needing a documented safety posture for regulated deployments, or for teams that must self-host the video model, since the open weights published so far cover FLUX 3 Image Klein and FLUX 3 Action. Competing shipped video models with longer track records and independent evaluation, such as Google's Veo 3.1 and OpenAI's Sora 2, are worth comparing directly, alongside Kling 3.0, Seedance 2.5, Wan 3.0 and the video tools Runway, Pika and Luma Dream Machine, since FLUX 3's own preference-test wins have not yet been reproduced by a third party.

Pricing

Video generation is billed per second, not per token, with rates by resolution class (read from Black Forest Labs' pricing data on 2026-10-18; 1 credit = $0.01). Text or image to video: $0.15/s HD, $0.24/s Full HD, $0.28/s QHD, $0.41/s UHD (4K). Video continuation: $0.21/s, $0.27/s, $0.31/s, $0.45/s.

Draft (HD): $0.06/s for text or image input, no continuation draft rate listed. Pay-as-you-go, no subscription. FLUX 3 Image is priced per image from $0.041 (768 px square) to $0.607 (4K), and FLUX 3 Image Klein from $0.008 to $0.036 per image.

Key Features

• Twenty-second single-pass video generation: Generates continuous clips twenty seconds long in one request, at resolutions from HD (720p) up to 4K, longer than most shipped competitors' single-generation cap.
• Native multilingual audio: Generates lip-synced dialogue in multiple languages, sound effects, and environmental ambience inside the same pass as the video, not as a separate dubbing step.
• Keyframe-controlled transitions: Accepts one start frame, start and end frames, or up to ten keyframes, optionally pinned to exact seconds, to control scene changes and camera moves within one generation.
• Video-to-video and continuation: Extends or transforms an existing clip, not just text/image-to-video, letting a workflow build on prior generations.
• Self-Flow shared architecture: One backbone underlies the family, including FLUX 3 Action, which predicts robot actions from camera images, state and an instruction.
• Draft mode with enhance: Drafts are cheap previews, and draft_enhance turns the one you pick into a full-quality render of the identical shot.
• FLUX Video Edit [fast]: A separate endpoint applies a text instruction to an existing clip and keeps the source clip's length.

Pros

• Longer single-generation video length than most shipped competitors' per-generation cap.
• Native audio (dialogue, effects, ambience) is generated with the video, removing a separate TTS/dubbing pass from the pipeline.
• Pay-as-you-go per-second pricing with a cheap Draft rate for fast iteration before a final render at HD, Full HD, QHD or 4K.

Cons

• Vendor documentation labels FLUX 3 a preview model, so features and rates can still change, and Omni Reference with images and videos is listed as coming soon.
• No independent third-party benchmark exists; the only quality comparisons are vendor-reported from an early checkpoint.
• No public system card or safety benchmark, unlike the documentation Black Forest Labs ships for its image models.

Visit FLUX 3 Official Page