# FLUX 3

Black Forest Labs, released 2026-07-23

**Verdict: conditional.** FLUX 3 is Black Forest Labs' unified multimodal model for image, video, audio and robot-action prediction. Its Video component is now pay-as-you-go through BFL's dashboard and API, from $0.06 per second at Draft HD to $0.80 at UHD, with five seconds of Standard HD costing $0.85. Conditional because every preference result is BFL's own, Image and Audio are not yet released, and the open-weight FLUX 3 Dev is still listed as coming soon. Last verified 2026-07-27.

## Pricing

- Input: $0.17 / sec (Standard HD text-to-video)
- Output: Included in per-second video rate
- Context: Video GA pay-as-you-go; $0.06-$0.80/s by res

## Strengths

- Unified multimodal generation — video, audio, images, and action prediction from one shared architecture trained jointly across all modalities
- Pay-as-you-go video through the BFL dashboard and API: text/image-to-video from $0.06/sec at Draft HD to $0.80/sec at UHD, with five seconds of Standard HD costing $0.85
- Native synchronized audio in video generation — dialogue, SFX, and music generated in the same pass as the video, not bolted on later
- Robotics action prediction via FLUX-mimic — deployed at Audi for soft-body manipulation tasks with ~10x sample efficiency over prior approaches
- Image or audio generation today: BFL lists FLUX 3 Image and Audio as separate, later parts of the rollout; only Video is on pay-as-you-go
- Open-weight deployment today — BFL lists the FLUX 3 Dev open weights as 'coming soon'
- Text or code generation — this is a visual intelligence model for creative media and robotics, not an LLM

## Benchmarks

- Human Preference vs Luma Ray 3.2 (10s 720p TTW): 93% (https://bfl.ai/blog/flux-3)
- Human Preference vs Runway Gen-4.5 (10s 720p TTW): 77% (https://bfl.ai/blog/flux-3)
- Human Preference vs Grok Imagine Video (10s 720p TTW): 69% (https://bfl.ai/blog/flux-3)
- Human Preference vs Kling v3 Pro (10s 720p TTW): 60% (https://bfl.ai/blog/flux-3)
- Human Preference vs Seedance 2.0 (10s 720p TTW): 52% (https://bfl.ai/blog/flux-3)
- Human Preference vs Gemini Omni Flash (10s 720p TTW): 52% (https://bfl.ai/blog/flux-3)

## Sources

- [Black Forest Labs: FLUX 3 model page (pricing and availability)](https://bfl.ai/models/flux-3)
- [MarkTechPost — Michal Sutter](https://www.marktechpost.com/2026/07/26/black-forest-labs-releases-flux-3-a-multimodal-flow-model-for-image-video-audio-and-robot-action-prediction/)
- [VentureBeat — Carl Franzen](https://venturebeat.com/technology/black-forest-labs-launches-flux-3-capable-of-generating-images-and-20-second-video-with-audio-but-in-limited-release-to-start)
- [Black Forest Labs Official Press Release (GlobeNewswire)](https://www.globenewswire.com/news-release/2026/07/23/3332364/0/en/black-forest-labs-unveils-flux-3-a-new-multimodal-frontier-model-for-visual-intelligence.html)
- [BFL Official Blog](https://bfl.ai/blog/flux-3)
- [BFL FLUX-mimic Technical Blog](https://bfl.ai/blog/flux-3-mimic)

Canonical: https://neomanex.com/models/flux-3