FLUX 3
bfl/flux-3FLUX 3 is Black Forest Labs' upcoming unified multimodal foundation that brings image, video, audio, language, and action prediction into one model. The Flux 3 API will cover text-to-video with native audio, image synthesis and editing, and physical-AI capabilities as each rolls out.
What Makes the Upcoming Flux 3 API a Major Leap
FLUX 3 is built around a single foundation for image, video, audio, language, and action prediction. These capabilities reflect what Black Forest Labs has announced and may be refined before wider release.
One Shared Multimodal Foundation
Learns from images, video, audio, language, and action data together, so FLUX 3 understands how a scene develops, which sounds belong to events, and how objects respond to motion.
Video and Native Audio Together
Designed to create visuals and audio in the same generation — clips up to 20 seconds with dialogue, ambience, and impacts aligned to on-screen movement.
Multiple Input Guidance
Beyond text, use images, keyframes, reference clips, and existing video-audio to shape the output and control transitions.
Longer, Consistent Stories
Chain separate generations into extended multi-shot sequences, carrying characters, objects, environments, and style across scenes.
Multilingual Dialogue and Visual Text
Place readable signs, titles, labels, and animated text inside images and videos, with multilingual dialogue for different audiences.
Physical Dynamics and Action
Learning from movement, sound, and object interaction supports more coherent physical behavior and a foundation for simulation and robotic manipulation.
FLUX 3 Video Showcase
Clips generated by the FLUX 3 API. Enable sound to hear native multilingual audio.
FLUX 3 use cases
FLUX 3 targets high-motion video — cinematic clips, ads, and VFX-style shots where prompt adherence and motion realism matter.
Cinematic clips
Filmic shots with camera motion and depth that hold up at 2K.
Ad spots
Prompt-driven ad clips with strong prompt adherence — what's described is what's rendered.
VFX-style shots
Stylized motion sequences for intros, transitions, and title cards.
Concept exploration
Spin many short variants to find a look before committing to a full render.
FLUX 3 vs Today's Leading Video Models
Preference data from Black Forest Labs' FLUX 3 announcement: 10-second, 720p text-to-video clips with native audio, head-to-head against competing models.
| Competing Model | FLUX 3 Preference Rate | What It Shows |
|---|---|---|
| Luma Ray 3.2 | 93% | Largest reported lead |
| Runway Gen-4.5 | 77% | Strong advantage |
| Grok Imagine Video | 69% | Clear advantage |
| Kling v3 Pro | 60% | Meaningful advantage |
| Seedance 2.0 | 52% | Closely matched |
| Gemini Omni Flash | 52% | Closely matched |
Figures reflect a preliminary internal evaluation of a development-stage candidate, not an independently reproduced benchmark. Sample size and methodology have not been published; treat as an early indication, not final results.
FLUX 3 API
Send a POST request to /api/v1/runs with your API key and input. Generation runs asynchronously — poll the run ID for status, then read the output URLs when it succeeds.
curl -s https://runbase.net/api/v1/runs \
-H "Authorization: Bearer $RUNBASE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "bfl/flux-3",
"input": { "prompt": "A cinematic product photo of a ceramic lamp" }
}'| Parameter | Type | Required | Description |
|---|---|---|---|
| prompt | string | Required | Prompt |
| aspect_ratio | string | Optional | Aspect ratioOptions: 1:13:44:39:1616:921:9 |
| duration | number | Optional | Duration (seconds) |
How to Prepare for the Flux 3 API
FLUX 3 is coming soon. Get ready now so you can ship the day access opens.
- 1
Create Your Runbase Account
Register or sign in to manage API credentials, balance, usage, and the models available through Runbase.
- 2
Generate an API Key and Prep Integration
Create a key and store it securely. Authentication, async task handling, callbacks, and media storage can all be prepared using the existing integration structure.
- 3
Select Flux 3 Once Available
After official support, review the latest docs and pick the capability your project needs — video and audio, image, or other phased-rollout features.
- 4
Configure and Submit Requests
Provide a text prompt plus any supported references (images, video, audio, keyframes), choose settings, and submit through the confirmed endpoint.
- 5
Track the Task and Retrieve Results
Use the returned task info to check status or receive a callback, then retrieve the finished media and usage details per the latest documentation.