Industry News
Black Forest Labs released FLUX 3 Video Part 1 on August 4, 2026. Here is what the generation capabilities include, how it compares to Veo 3.1 and Runway, and what is missing from the roadmap.
Some links are partner links: if you subscribe through them, we may earn a commission, at no extra cost to you. The crowd verdicts stay independent.
FLUX 3 Video Part 1 launched on August 4, 2026. Black Forest Labs, the team behind FLUX.2 (71% crowd approval in the image arena), shipped generation capabilities for text-to-video, image-to-video and video continuation, with native audio and clips up to 20 seconds.
The headline numbers
FLUX 3 Video generates HD clips up to 20 seconds with native audio. Early testers preferred it over Runway Gen-4.5 in 77% of head-to-heads and over Luma Ray 3.2 in 93%. No pricing yet.
What Part 1 generation actually includes #
The Part 1 release covers five input modes:
Text-to-video: describe a scene in plain language; the model handles motion, scene logic and audio together.** Image-to-video and keyframes**: start from a still image, specify an end frame, or set multiple keyframes for controlled transitions.** Video continuation**: extend an existing 4-second clip with coherent audio and visuals.** Multiple shots**: generate varied scenes and camera angles within a single prompt.
Audio is baked into the video generation, not added as a post-process. The model produces dialogue in English, Chinese, Spanish, French, German, Japanese, Portuguese, Russian, Italian, Indonesian, Turkish, Hindi and Punjabi with lip-synced output.
Technical specs at launch #
| Spec | Value |
|---|---|
| Max duration | 20 seconds |
| Resolution at launch | 720p (1080p rolling out) |
| Native audio | Yes (dialogue, SFX, ambient) |
| Input modes | text, image, video, audio |
| Reference images | up to 10 |
Black Forest Labs also added a Draft Mode for fast iteration at lower cost before committing to a full render.
How it stacks up against competitors #
The AI video field in August 2026 has clear leaders. Here is how FLUX 3 compares to what is already shipping:
| Model | From | Resolution | Max duration | Native audio |
|---|---|---|---|---|
| FLUX 3 Video | TBA | 720p (1080p soon) | 20s | Yes |
| Google Veo 3.1 | $20/mo | 1080p | 8s (148s extended) | Yes (48kHz spatial) |
| Runway Gen-4.5 | $12/mo | 1080p | 10s | Yes |
| Seedance 2.0 | $15/mo | 2K | 15s (30s on 2.5) | Yes |
| Kling 3.0 | $7/mo | 4K | 15s | Yes |
In early blind tests, FLUX 3 beat Runway Gen-4.5 77% of the time and Luma Ray 3.2 93% of the time. It ties Seedance 2.0 in image-to-video and "surpasses other alternatives," per the official announcement.
The 20-second clip ceiling matters. Kling AI caps at 15 seconds, Veo 3.1 at 8 seconds per generation. For single-generation length, FLUX 3 is already competitive, though Scene Extension on Veo chains clips up to 148 seconds.
FLUX
The open-weight champion: 32B FLUX.2 wins 66.6% of head-to-heads vs open rivals, API from $0.014 per image
Partner link. The crowd verdicts stay independent.
What is missing from Part 1 #
The August release is generation only. The FLUX 3 roadmap promised four pillars: image, video, audio and action (robotics). Part 1 ships video. What is not here yet:
FLUX 3 Image: editing and generation, "coming weeks" per the blog post.** FLUX 3 Dev (open weights): planned for later in 2026, no date locked. Pricing**: the BFL pricing page lists FLUX.2 and FLUX Tools, but FLUX 3 has no row yet.
Access today is through the BFL API and select partners. The consumer web app and open weights are downstream.
The pricing gap #
Black Forest Labs has not published FLUX 3 pricing. For context, here is what FLUX.2 costs:
- FLUX.2 [klein]: $0.014/image (4B) to $0.015/image (9B)
- FLUX.2 [pro]: $0.03/image
- FLUX.2 [max]: $0.07/image (first megapixel)
About $3 covers 100 [pro] images. Video will cost more per second than images cost per frame, but the BFL team has historically undercut competitors on price. The 71% crowd approval for FLUX.2 in the AI image leaderboard suggests early trust.
FLUX
The open-weight champion: 32B FLUX.2 wins 66.6% of head-to-heads vs open rivals, API from $0.014 per image
Partner link. The crowd verdicts stay independent.
The verdict #
Part 1 is real: 20-second clips with native audio, 5 input modes, multilingual lip sync.** Performance is strong**: 77% win rate vs Runway Gen-4.5 in blind tests, 93% vs Luma Ray 3.2.** Gaps remain**: no pricing, 720p at launch, no open weights until later in 2026.** For FLUX.2 users**: the image model (71% recommend, $3/100 images) is the entry point. FLUX 3 Video builds on that foundation.
The August release positions Black Forest Labs as a serious contender against Veo, Runway and Seedance. Pricing and open weights will determine whether that translates to adoption. Check the AI video leaderboard for live crowd scores as the rollout continues.
Keep exploring
Every claim above is backed by the arena's live data: crowd votes, verified pricing, honest pros & cons.
More from the arena journal #
Industry News
FLUX 3 Drops: Black Forest Labs Merges Image, Video and Audio Into One Model
Black Forest Labs releases FLUX 3, the first multimodal model generating images, 20-second video with native audio, and robot actions from one architecture.
Aug 3, 2026 · 4 min read
Industry News
Can You Legally Sell AI Music From Suno and Udio in 2026
Complete breakdown of commercial licensing terms, ownership rights and monetization paths for AI-generated music across Suno, Udio, ElevenLabs Music and Stable Audio.
Aug 7, 2026 · 5 min read
Industry News
DeepSeek V4 vs Claude Opus 4.8: Can the $0.87 Model Compete in Agentic Coding?
DeepSeek V4 costs 28.7x less than Claude Opus 4.8 per output token. Both score 57% crowd approval, but the real tradeoff lies in tool-call reliability vs raw price. Here is what the data shows.
Jul 27, 2026 · 4 min read