Home FLUX 3

New in Renderforest: Black Forest Labs' FLUX 3

FLUX 3 Video AI Video Generator with audio, up to 20 seconds in one take

Black Forest Labs' FLUX 3 turns a prompt or a photo into a 5- to 20-⁠second clip with dialogue, sound effects and ambience generated with the picture. Set a start and an end frame, then cut, caption and publish in the Renderforest editor. No API key, no GPU.

Try FLUX 3

Dialogue, sound effects and ambience in one take · Draft first with Renderforest Fast, then finish on FLUX 3

ModelFLUX 3Also on your plan: Gemini Omni Flash, Kling 3.0, Seedance 2.5, MiniMax H3, Pixverse V6
Add a prompt to generate your video
  • 4.8on Capterra
  • 4.7on G2
  • 100+countries
  • 40M+users
  • 120M+projects since 2013

Prompt to published in four steps

FLUX 3 is Black Forest Labs' multimodal model for video with native audio, announced on July 23, 2026 and generally available since August 4, 2026. It generates 5- to 20-second clips with synchronized dialogue, sound effects and ambience from text, images or keyframes. In Renderforest you run it in a browser, then finish the video in the editor.

  1. Step 1: Choose FLUX 3

    Open the AI video generator and pick FLUX 3 in the model list. Add a photo if the shot should start from it.

  2. Step 2: Prompt like a director

    Describe the shot, the camera and the sound. Put spoken lines and on-screen words in quotes, and name the language if it isn't English.

  3. Step 3: Set length, frames and format

    Choose 5 to 20 seconds, 720p or 1080p, the aspect ratio and an optional end frame.

  4. Step 4: Edit and publish

    The clip lands in the editor: trim, caption, add music or a voice-over, apply your brand kit and export for every platform.

What FLUX 3 does here

One take, up to 20 seconds

Generate 5 to 20 seconds in a single pass, in any whole-second length: a full ad beat, a two-line dialogue or an establishing shot without stitching clips together.

Sound generated with the picture

Dialogue, sound effects and ambience come out of the model in sync with the frame. FLUX 3 speaks English, Spanish, French, German, Italian, Portuguese, Russian, Turkish, Japanese, Chinese, Indonesian, Hindi, Punjabi and more.

Start and end frames

Use a photo as the first frame, add a second as the last, and FLUX 3 builds the motion between them: product reveals, before-and-afters, seamless loops.

Past 20 seconds, and into the editor

Chain FLUX 3 scenes into a longer story with Long AI Video, or finish any clip in the editor with captions, music, your brand kit and one-click resize.

Several shots in one generation

Describe the cuts (“wide shot, then a close-up on her hands”) and FLUX 3 can stage more than one camera angle in a single clip.

Text that holds on screen

Put the exact words in quotes and FLUX 3 renders titles, signs and labels inside the scene, moving with it.

Black Forest Labs' FLUX 3: what it is, what's new, and what runs here

FLUX 3 is the third generation of FLUX, the model family from Black Forest Labs, the Freiburg lab founded in 2024 by the researchers behind Stable Diffusion. FLUX.1 and FLUX.2 made images; FLUX 3 is one multimodal model for video with native audio, for images and for robot actions. FLUX 3 Video, the video-with-audio part, is what runs in Renderforest.

Timeline

  1. FLUX.1: Black Forest Labs' first image models.

  2. FLUX.2: production image generation and editing.

  3. FLUX 3 announced: one model for video with audio, images and robot actions; early access.

  4. FLUX 3 Video generally available through BFL's API and partners: 5-20-second clips, 720p and 1080p, native audio.

  5. 2K and 4K output tiers added to BFL's API, finished by an upsampler.

  6. FLUX 3 Image (hosted) and FLUX 3 Image Klein (open weights, Oct 15); FLUX 3 Max, fal's faster post-trained variant of FLUX 3 Video.

  7. FLUX 3 Video in Renderforest's video generator.

FLUX 3 Video in the model and in Renderforest (checked Nov 11, 2026)
In Renderforest todayFLUX 3 Video (Black Forest Labs API)
Clip length5-20 s, any whole second5-20 s from text or images; 5-15 s when continuing a clip
Resolution720p and 1080p720p and 1080p; 2K and 4K through an upsampler
Frame rate24 fps24 fps
Aspect ratios16:9, 9:16, 1:121:9, 2:1, 16:9, 4:3, 1:1, 3:4, 9:16
AudioNative audio includedNative, on by default: dialogue, sound effects, ambience; speech in 13 named languages “and more”
Image inputStart frame and end frame1 image = start frame; 2 = start and end; up to 10 keyframes pinned to times
ReferencesNot yetImage and video references announced as “coming soon”
Extend a clipNot yetContinuation from up to the last 4 s of a clip
Draft modeNot yetFast 720p drafts, then enhance the keeper
Faster variantNot yetFLUX 3 Max (fal's post-trained variant, lower latency)

Built for the people shipping video weekly

Performance marketers

A 15-second ad with the spokesperson's line, the product sound and the end-card text in one take; brand kit on every cut.

Make an ad clip

Social & YouTube creators

Vertical clips with dialogue and ambience, captioned in the editor and posted the same day.

Make a Short

Ecommerce & product teams

Your product photo as the first frame, the hero angle as the last; FLUX 3 animates the reveal between them.

Animate a product photo

Global and localization teams

Spoken lines in Spanish, German, Japanese, Hindi and more, generated with the scene; add subtitles or translate the finished video.

Translate a video

Filmmakers & storyboarders

Multi-shot pre-vis in one generation; chain scenes past 20 seconds in Long AI Video.

Pre-vis a scene

Agencies

Every frontier video model in one place, with client brand kits.

Explore plans

Prompts in, campaigns out

  • Rainy market POV

    • 7 s
    • 720p
    • 16:9

    Prompt: “First-person handheld POV walking through a covered Japanese shopping arcade on a rainy afternoon. A glass roof overhead, rain pouring past the open end of the arcade, the wet tiled floor reflecting warm shop lights. On the left a fish stall with whole fish on ice under hanging bulbs; a motorbike rider in a black helmet cruises past, spraying water. The camera turns to an old secondhand bookshop crammed with stacked books, where an elderly shopkeeper in an apron and flat cap tends a small wood stove. My hands, in a mustard-yellow raincoat with grey sleeves, reach out, and he passes me a warm roasted sweet potato wrapped in brown paper while a fluffy black-and-white cat and a scruffy grey dog look up from beside the stove. Close-up as I break the sweet potato open over the wet street: golden, steaming flesh, rain splashing behind it. Sound: heavy rain drumming on the glass roof, a motorbike engine and a splash, the stove crackling, the shopkeeper saying softly in Japanese, “Dōzo,” paper rustling. Natural light, cosy documentary feel, 16:9.”

  • Dinosaur escape

    • 6 s
    • 720p
    • 16:9

    Prompt: “Cinematic survival thriller, photoreal, golden late-afternoon light in a prehistoric valley of ferns, cycads and dry grass. Open on an extreme low angle as a man's boot slams into a muddy puddle, water and mud exploding toward the lens. Cut to a tight handheld shot of his face, a young man with short brown hair and a dark sweater, glancing back in terror as a huge horned dinosaur, like a Triceratops, charges out of the dust behind him. A wide tracking shot as he vaults a fallen log and sprints along the dirt trail, the dinosaur crashing through the brush and the ground shaking. Final shot from behind as he dives toward a rocky arch and into the shadows, dust and leaves swirling. Fast, kinetic camera with motion blur, warm haze, shallow depth of field. Sound: pounding footsteps and heavy breathing, splashing mud, a deep bellowing roar, thundering steps and snapping branches, a tense percussive score. 16:9.”

  • Midnight guitar

    • 9 s
    • 720p
    • 16:9

    Prompt: “Intimate music video, moody low light. A young man with slicked-back dark hair, a small hoop earring and a black denim jacket plays a worn black electric guitar alone in a dark, empty room. Warm table lamps glow softly out of focus in the background, and a bare bulb behind him blooms into a gentle lens flare as the camera drifts. A slow, smooth orbit from a side profile to a three-quarter view, staying close on his face and hands; he keeps his eyes down on the fretboard, completely absorbed. Deep shadows, warm amber highlights on his face and on the guitar's chipped finish, subtle film grain, shallow depth of field, a 35mm look. Sound: a clean, reverb-soaked electric guitar riff, slow and melancholic, the soft buzz of the amp, fingers sliding on the strings, the quiet hum of the room. One continuous take, 16:9.”

  • Warehouse face-off

    • 8 s
    • 720p
    • 16:9

    Prompt: “Gritty action thriller in a dim warehouse lined with tall racks of shrink-wrapped pallets, shafts of cold light cutting through haze and dust. A huge bald, bearded man in a black tank top and camouflage trousers, a pistol holstered on his belt, looms over a slim young woman with a dark ponytail in a white zip-up hoodie, who calmly smokes a cigarette. A medium two-shot, then a tense close-up as he leans in and growls, “You shouldn't have come here.” She exhales smoke straight into his face. In one fluid move she spins and lands a high roundhouse kick to his chest, her hoodie flaring, the haze swirling around them. Handheld camera, cold blue-grey grade, hard backlight, sharp contrast. Sound: echoing warehouse ambience, the low hum of the lights, his gravelly voice, a long exhale, the snap of fabric and a heavy thud on impact, a pulsing bass score. 16:9.”

  • Golden-hour walk

    • 7 s
    • 720p
    • 16:9

    Prompt: “Golden-hour fashion film on a busy Manhattan avenue, the low sun shining straight down the street between skyscrapers, the Empire State Building silhouetted on the horizon, steam rising from street vents. A young woman with a short wavy bob, round dark sunglasses and over-ear headphones walks toward the camera through the crowd, cool and unhurried. She wears an oversized dark brown trench coat, a white shirt with a black tie, wide black trousers and a small crossbody bag, one hand in her pocket. The crowd streams past her in soft motion blur; she glances to the side, then back into the lens. A slow backward tracking shot at her pace, long shadows stretching across the asphalt, warm backlight and flare, rich contrast, a 50mm look. Sound: rush-hour city ambience, footsteps, distant horns and chatter, the hiss of steam, a muffled lo-fi beat leaking from her headphones. One continuous take, 16:9.”

  • Bobblehead portrait

    • 8 s
    • 720p
    • 16:9

    Prompt: “Playful, surreal portrait in a bright, airy art gallery with concrete walls and large black-and-white photo prints. A young woman with dark curly hair, a black corduroy cap, olive-green rectangular sunglasses and a beige trench coat over a navy sweater looks into the camera. Two hands enter the frame and squish her cheeks; when they let go, her head pops into a cartoonish, oversized bobblehead on a tiny body, wobbling gently, while the room stays completely real. She purses her lips, then breaks into a slow, smug little smile. A locked-off medium close-up, soft natural daylight from tall windows, clean pastel tones, crisp detail. Sound: quiet gallery ambience, a soft squish, a springy cartoon boing as her head wobbles, a short amused “hm”, light playful plucked strings. 16:9.”

FLUX 3, explained properly

  1. What is FLUX 3, and which FLUX 3 is which?

    FLUX 3 is Black Forest Labs' multimodal model family, announced on July 23, 2026: one model trained across video, audio, images and actions. FLUX 3 Video generates clips with native audio and is the model in Renderforest. FLUX 3 Max is fal's post-trained variant of FLUX 3 Video, tuned for prompt adherence and lower latency. FLUX 3 Image and FLUX 3 Image Klein (5.6B parameters, open weights) make images. FLUX 3 Action predicts robot actions with industrial partners. FLUX 3 Dev, an open-weights version of the backbone, has been announced without a date. FLUX 3 is not Flux.ai, the circuit-board design tool, and it is the first FLUX that makes video at all: FLUX.1 and FLUX.2 are image models.

  2. How to prompt FLUX 3 like a director

    FLUX 3 reads a prompt as a shot list. Lead with the subject and the action, then the camera (“handheld”, “slow push-in”, “one continuous shot”), the light, and the sound: name what should be heard (“rain on a tin roof, no music”). Put spoken lines in quotes and say the language (“she says in German: ‘Wir haben geöffnet.’”). Put on-screen text in quotes too. For several shots, describe each cut in order. With start and end frames, describe the motion between them, not the frames themselves. Keep negatives in plain words inside the prompt. Draft the idea on Renderforest Fast, then run the keeper on FLUX 3.

  3. FLUX 3 vs Seedance 2.5, Gemini Omni Flash, Kling 3.0 and MiniMax H3

    In Black Forest Labs' own preliminary test (July 2026; 10-second 720p clips with audio, human raters), FLUX 3 was preferred over Kling v3 Pro 60% of the time and over Seedance 2.0 and Gemini Omni Flash 52% of the time, close to a tie. On the independent Design Arena leaderboard (Aug 7, 2026) it ranked second of 39 video models, behind Gemini Omni Flash. Inside Renderforest: pick FLUX 3 for 10-20-second single takes where dialogue, sound and on-screen text must land together; Seedance 2.5 for the longest takes (up to 30 s) and many references; Gemini Omni Flash for 4K finals and up to 10 references; Kling 3.0 for 4K and complex motion; MiniMax H3 for 1440p; Renderforest Fast to draft free, then re-render the keeper.

  4. Where FLUX 3 runs: Black Forest Labs' API, partner apps, and here

    Black Forest Labs offers FLUX 3 Video through its API and developer dashboard, plus a playground for developers. Partners resell it: fal (which also offers FLUX 3 Max), ElevenLabs, Picsart and others. The FLUX 3 Video weights are not public, so it does not run on your own GPU; the open-weights FLUX 3 Dev has been announced without a date, and the open FLUX 3 Image Klein makes images, not video. In Renderforest FLUX 3 runs in the browser, and from Claude through the Renderforest MCP app.

  5. Rights, safety and what you can publish

    Videos you make with FLUX 3 on a paid Renderforest plan can be used commercially, and Renderforest adds no visible watermark on paid plans. Black Forest Labs evaluated FLUX 3 Video with a third-party safety partner, Cinder, for risks including non-consensual intimate imagery and child sexual abuse material before release. Renderforest's content policy applies on top: no sexual content, no realistic depictions of real people without consent, no deceptive political content. Platforms such as YouTube, TikTok and Meta ask you to label realistic AI video; follow their rules when you post.

Safe to publish, ready for commercial use

Every prompt and upload is checked automatically before anything is generated. Renderforest does not create deceptive content or videos of real people without their consent, so what you publish stays safe for your brand, your clients and your channels. Read our Acceptable Use Policy.

Loved by Creators

  • ★★★★★

    “I personally use Renderforest to create video lessons for my online educational courses, and it has made the whole process much easier for me. I don’t come from a video production background, but the platform allows me to turn my scripts into clear, professional videos without any hassle. What I like most is how fast and straightforward everything is.”

    Kashif SyedTrustpilot
  • ★★★★★

    “…as a no code tech professional, drag and drop platforms are my bread and butter. With renderforest I'm able to create awesome motion graphics and explainer videos for my clients that look like Netflix!”

    Daniel R. S.Capterra
  • ★★★★★

    “Renderforest's AI-powered creative tools helped us transform complex psychological and emotional themes into visual storytelling. … What made a meaningful difference was the professionalism, kindness, and willingness of the Renderforest support team to listen, investigate difficulties, and help us find practical solutions.”

    Milorad CakardicTrustpilot
  • ★★★★★

    “Renderforest is easy to use and I have enjoyed the simplicity it brings to creating professional videos. My friends think I am a video guru because of Renderforest. … I have also used Renderforest to create animations and explainer videos for my YouTube channel.”

    Likhwa M.Capterra
  • ★★★★★

    “Awesome features. Short learning curve and very intuitive. I could create a product explainer video without any knowledge of video editing in just a couple hours of signing up for Renderforest. … This is a blessing in disguise for startups”

    Teja GudluruTrustpilot
  • ★★★★★

    “I love that Renderforest is simple to use and offers a variety of services. I have used this product for many years, and it just keeps getting better. … As an author, I use the program to create book trailer videos. I also use it to create marketing videos for others.”

    Latanya B.Capterra
  • ★★★★★

    “In managing various video, music, and multimedia productions, efficiency is essential, which is why Renderforest has been a go-to platform for me since 2022. The user-friendly interface and versatile video templates integrate seamlessly into my project pipelines, whether I am utilizing pre-built assets or constructing visuals from the ground up.”

    Black ShadowTrustpilot
Frequently asked questions

FLUX 3 questions, answered straight

What is FLUX 3, and who makes it?

FLUX 3 is a multimodal AI model from Black Forest Labs, the German lab behind the FLUX image models. Announced on July 23, 2026, it generates video with native audio, images and robot actions from one model. FLUX 3 Video became generally available on August 4, 2026, and you can use it in Renderforest's AI video generator.

Is FLUX 3 a video model or an image model?

Both, as separate products. FLUX 3 Video makes 5- to 20-second clips with sound: that is the model on this page. FLUX 3 Image and the open-weights FLUX 3 Image Klein make still images. For images in Renderforest, use the AI Image Generator, which offers FLUX image models among others.

Is FLUX 3 free?

Not through official channels: Black Forest Labs offers FLUX 3 Video through its API for developers, and the video weights are not public. In Renderforest, FLUX 3 comes with the paid plans; on the free plan you can draft with Renderforest's own model and upgrade when you want FLUX 3.

How long can FLUX 3 videos be?

FLUX 3 generates 5 to 20 seconds in a single take, in any whole-second length. In Renderforest you choose the length before you generate; for anything longer, chain scenes in Long AI Video or join clips in the editor.

What resolution and aspect ratios does FLUX 3 support?

Renderforest renders FLUX 3 at 720p or 1080p in 16:9, 9:16 and 1:1. Black Forest Labs' API also offers 2K and 4K, finished by an upsampler, and wider ratios such as 21:9 and 4:3.

Does FLUX 3 generate audio, dialogue and lip sync? In which languages?

Yes. FLUX 3 generates dialogue, sound effects and ambience with the picture, with lip-synced speech. Black Forest Labs names English in several dialects, Chinese, Spanish, French, German, Japanese, Portuguese, Russian, Italian, Indonesian, Turkish, Hindi and Punjabi, “and more”. Put spoken lines in quotes and name the language.

Can I start from my own image or set start and end frames?

Yes. Upload a photo as the first frame and FLUX 3 animates from it; add a second image as the last frame and it builds the motion between the two. Black Forest Labs' API also accepts up to ten keyframes pinned to times; Renderforest exposes the start and end frame today.

What is FLUX 3 Max, and can I use it here?

FLUX 3 Max is fal's post-trained variant of FLUX 3 Video, tuned for stronger prompt adherence and lower latency. It runs on fal and its partners. Renderforest runs FLUX 3 Video today; FLUX 3 Max is not in the model list yet.

Is FLUX 3 open source? Can I run it locally?

Not the video model. The FLUX 3 Video weights are not public, so ComfyUI or local setups use it through Black Forest Labs' API. Black Forest Labs has announced an open-weights FLUX 3 Dev without a date, and FLUX 3 Image Klein is open but makes images. In Renderforest FLUX 3 runs in the browser, with no GPU.

FLUX 3, Seedance 2.5 or Gemini Omni Flash: which should I pick?

FLUX 3 for 10-20-second single takes with dialogue, sound and on-screen text; Seedance 2.5 for takes up to 30 seconds and many references; Gemini Omni Flash for clips up to 10 seconds, and for 4K finals. All three are in the same picker; draft on Renderforest Fast, then re-render the keeper.

Is there a FLUX 3 API?

Yes. Black Forest Labs sells FLUX 3 Video through its API for developers. You do not need it in Renderforest: pick FLUX 3 in the model list. You can also generate FLUX 3 clips from Claude through the Renderforest MCP app.

Is FLUX 3 the same as Flux AI or Flux.ai?

FLUX 3 is the newest model in Black Forest Labs' FLUX family, which people often call “Flux AI”. It is not related to Flux.ai, a circuit-board design tool that shares the name. Renderforest is not affiliated with Black Forest Labs; FLUX is their trademark.

Can I use FLUX 3 videos commercially? Do they have a watermark?

Yes, on paid plans: videos you make with FLUX 3 in Renderforest can be used in commercial projects, and Renderforest adds no visible watermark on paid plans. Platforms may ask you to label realistic AI-generated video when you post it.

What kind of content is not allowed?

Renderforest's content policy and Black Forest Labs' safety filters block sexual content, content involving minors, realistic depictions of real people without consent and deceptive or violent content. Prompts that break the policy are refused. Read our Acceptable Use Policy.

Stop rationing renders. Start shipping videos.

One place for every model. Unpredictable creativity.

FLUX 3 with native audio, every other frontier model, the full editor, templates and your brand kit, all in one place.

  • Every model on one plan
  • New models as they launch
  • The full editor included