T2VLive
Live endpointsFLUX 3 Text to Video
Cinematic text-to-video with optional native synchronized audio and physically grounded motion.
720p–1080p
$0.25–$0.42/secVideo API Live11 endpoints
FLUX 3 API — Video, Image & Audio API by Black Forest Labs
FLUX 3 is Black Forest Labs' unified multimodal model family, jointly trained across image, video, and audio. Its live video endpoints add physically grounded motion and optional native synchronized audio to the same integration.FLUX 3 image, Dev, and Omni variants are coming soon. Muapi will activate each endpoint as general availability opens, without requiring a second signup.
Live FLUX 3 endpointsVideo generation, extension, drafts, and upscaling7 endpoints
Try Model
T2VLive
Live endpointsFLUX 3 Text to Video Draft
Fast, lower-cost draft mode for ideation, storyboarding, and prompt iteration before a final render.
720pFast
$0.09/secI2VLive
Live endpointsFLUX 3 Image to Video
Animate a still image into cinematic video while keeping motion physically consistent with the source frame.
720p–1080p
$0.25–$0.42/secI2VLive
Live endpointsFLUX 3 Start-End to Video
Generate a controlled transition between a supplied start and end keyframe, with optional native audio.
720p–1080p
$0.25–$0.42/secExtendLive
Live endpointsFLUX 3 Video Extend
Continue an existing video with prompt-guided motion, scene development, and optional native synchronized audio.
720p–1080p
$0.25–$0.42/secExtendLive
Live endpointsFLUX 3 Video Extend Draft
Fast, lower-cost continuity testing before extending a clip with the final-quality Video Extend endpoint.
720pFast
$0.09/secUpscaleLive
Live endpointsFLUX 3 Video Upscaler
Upscale FLUX 3 or other video output beyond native resolution while preserving motion coherence and audio sync.
Up to 4×
$1.43–$2.00/runComing soonImage, Dev, and Omni Reference variants4 endpoints
T2ISoon
Coming soonFLUX 3 Text to Image
Flagship text-to-image generation with sharper detail and stronger prompt adherence than earlier FLUX image models. Coming soon.
Up to 4K
TBDEditSoon
Coming soonFLUX 3 Image to Image
Instruction-driven image editing with up to four reference images, built on FLUX 3's unified architecture. Coming soon.
Up to 4K
TBDT2ISoon
Coming soonFLUX 3 Dev
Faster, lower-cost FLUX 3 variant planned for open-weight release later in 2026. Coming soon.
Up to 4KFast
TBDOmniSoon
Coming soonFLUX 3 Omni
Multi-reference image and video generation that keeps subject, style, and scene consistent across outputs. Coming soon.
Up to 4K
TBDWhat is FLUX 3?
FLUX 3 is Black Forest Labs' newest frontier model, announced in July 2026 as a unified multimodal system jointly trained across image, video, and audio data. It extends the same architecture toward action prediction, giving the family a stronger grasp of real-world lighting, motion, and physics than image-only models such as FLUX Kontext.
Muapi's FLUX 3 video API is live now. Text to Video, Image to Video, Start-End to Video, Video Extend, draft modes, and Video Upscaler share a consistent async REST lifecycle; the image, Dev, and Omni variants will join as they open.
Key Capabilities
Unified image, video, and audio
One architecture jointly trained across image, video, and audio data, with a path toward action prediction for robotics.
Native synchronized audio
Live FLUX 3 Video endpoints can generate scene-appropriate audio synchronized to the clip without a separate TTS step.
Physically grounded motion
Motion, lighting, and object behavior stay consistent with real-world physics across generated clips.
Instruction-driven editing
The planned FLUX 3 Image-to-Image endpoint accepts up to four reference images plus a text instruction.
Fast Dev variant
The planned Dev release trades some peak fidelity for lower latency and cost, with open weights expected later in 2026.
Up to 4K image output
FLUX 3 Image is expected to support output up to 4K, continuing the high-resolution direction of the FLUX family.
Video extension
Video Extend continues an existing clip with prompt-guided motion and scene development.
Start-end keyframes
Start-End to Video controls the transition between two supplied frames.
Video upscaling
Video Upscaler raises video beyond native resolution while preserving motion coherence and audio sync.
Fast draft modes
Text to Video Draft and Video Extend Draft lower cost and latency during prompt iteration.
Omni Reference
The planned Omni mode uses multiple image and video references to hold a subject, style, and scene consistent.
FLUX 3 Variants on Muapi
| Variant | Modality | Resolution | Best For | Status |
|---|---|---|---|---|
| FLUX 3 Text to Image | Image | Up to 4K | Flagship image generation | Soon |
| FLUX 3 Text to Video | Video + Audio | 720p–1080p | Cinematic clips with native audio | Live |
| FLUX 3 Image to Video | Video + Audio | 720p–1080p | Physically consistent image animation | Live |
| FLUX 3 Video Extend | Video + Audio | 720p–1080p | Continue an existing clip | Live |
| FLUX 3 Omni | Image + Video | Up to 4K | Multi-reference consistency | Soon |
FLUX 3 Video API
Submit a generation request with your Muapi API key, store the returned request ID, and poll the standard prediction result endpoint. The same lifecycle works across the live FLUX 3 video endpoints.
Submit a video generation task
curl -X POST https://api.muapi.ai/api/v1/flux-3-text-to-video \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt":"A cinematic aerial shot of mist moving through a mountain valley at sunrise.","resolution":"720p","duration":5}'
# Response: {"request_id":"REQUEST_ID"}Choose Text to Video, Image to Video, Start-End to Video, or Video Extend and send the endpoint-specific input fields.
Retrieve the completed result
curl https://api.muapi.ai/api/v1/predictions/REQUEST_ID/result \ -H "x-api-key: YOUR_API_KEY" # When status is completed, read the generated media URL from outputs.
Poll until status is completed, then read the generated media URL from the response.
Frequently Asked Questions
What is FLUX 3?
FLUX 3 is Black Forest Labs' multimodal frontier model, jointly trained across image, video, and audio within a unified architecture and extendable to action prediction.
Is the FLUX 3 API available right now?
The FLUX 3 video API is live on Muapi, including Text to Video, Image to Video, Start-End to Video, Video Extend, Video Upscaler, and fast draft modes. Image, Dev, and Omni variants are coming soon.
What FLUX 3 variants does Muapi offer?
Muapi lists eleven endpoints: seven live video-oriented endpoints and four coming-soon image, Dev, and Omni variants.
How is FLUX 3 different from FLUX.2 and FLUX Kontext?
FLUX.2 and FLUX Kontext focus on images. FLUX 3 unifies image, video, and audio generation, adding native synchronized audio and stronger real-world motion coherence.
Will FLUX 3 have an open-weight release?
Black Forest Labs has stated that faster open-weight versions will follow later in 2026. The Dev variant is the planned lower-cost path.
What is FLUX 3 Omni?
Omni Reference is the planned multi-reference mode for generating images or video from multiple image and video references while preserving subject and style consistency.
How much does the FLUX 3 video API cost?
Live main video endpoints bill $0.25 per second at 720p and $0.42 per second at 1080p. Draft modes bill $0.09 per second, while Video Upscaler ranges from $1.43 to $2.00 per run.
How do I start using FLUX 3 through Muapi?
Create a Muapi account and API key, then call any live endpoint. Coming-soon endpoints will activate for existing users as they become available.
Use FLUX Kontext today →Generate and edit images while FLUX 3 image variants roll out.Browse video models →Compare live text-to-video and image-to-video endpoints.Get FLUX 3 access →Create one Muapi API key for the live endpoints.
Ready to build with FLUX 3?
Start with the live video endpoints and keep one integration ready for the image and Omni releases.