Black Forest Labs Announces FLUX 3 Video, First Details Cover Generation Capabilities
Black Forest Labs has published the first part of its FLUX 3 Video release notes, focused on the model's generation capabilities. Full technical specifications, pricing, and benchmark data have not yet been disclosed.
Black Forest Labs has begun rolling out documentation for FLUX 3 Video, a new video generation model, starting with a post titled "FLUX 3 Video, Part 1: Generation." The release marks the company's continued expansion beyond image generation, where its FLUX.1 family has become a widely used open and commercial alternative to models from Stability AI and Midjourney.
As of publication, Black Forest Labs has not released full technical specifications for FLUX 3 Video. No context window size, parameter count, pricing per 1M tokens, or independent benchmark scores have been disclosed. The "Part 1" framing suggests this is the opening entry in a multi-part rollout, with additional posts likely to cover editing, control, or other capabilities in subsequent releases — following a pattern the company has used for prior FLUX launches.
What is confirmed
The only verified fact at this stage is the existence of a model called FLUX 3 Video and that Black Forest Labs is treating "generation" as a distinct, dedicated capability worth its own documentation. This implies the model supports text-to-video or image-to-video synthesis, consistent with the broader video generation market that includes OpenAI's Sora, Google DeepMind's Veo, and Runway's Gen series.
Beyond this, specific claims about output resolution, video length, frame rate, or generation speed have not been provided in the source material. Any performance figures Black Forest Labs shares in follow-up posts should be treated as company claims until independently verified by third parties.
Context: FLUX's trajectory
Black Forest Labs, founded by former Stability AI researchers, built its reputation on the FLUX.1 image models, which shipped in multiple tiers (schnell, dev, pro) with varying licensing terms and inference costs. A jump to "FLUX 3" alongside a dedicated video variant indicates the company is consolidating its next-generation architecture across both image and video modalities rather than treating video as a separate product line.
The absence of pricing and benchmark data at this stage is typical for staged releases, where technical reports precede API availability or open-weight downloads. Readers should expect Black Forest Labs to publish inference cost, hardware requirements, and comparative benchmarks (likely against Sora, Veo, or Kling) in later parts of this series.
What this means
This is a documentation-stage announcement, not a full product launch. There is no way yet to evaluate FLUX 3 Video's actual quality, cost, or availability against competitors. The practical significance will depend entirely on what Black Forest Labs discloses in the promised follow-up parts — particularly whether the model is open-weight (as FLUX.1 dev and schnell were) or closed and API-only, and how its pricing compares to existing video generation services. Until then, this announcement functions primarily as a signal that Black Forest Labs is moving into video generation as a core product line rather than a side experiment.
Related Articles
TBC launches Mischief, a Wan 2.2-based image-to-video model, on fal.ai October 30
The Biological Computing Co. (TBC) is launching TBC Mischief, an image-to-video model based on Wan 2.2, through fal.ai on October 30. TBC says it generates video five times faster at 80% lower cost than the underlying model, while improving output quality.
NVIDIA and MIT team releases Open-H3-Regenerate-2K, regenerating MiniMax H3 video drafts at up to native 4K
The Efficient-Large-Model team, made up of NVIDIA and MIT researchers, released Open-H3-Regenerate-2K weights on Hugging Face. Built on MiniMax H3, it regenerates low-resolution video drafts at 2560×1440 and up to native 3840×2160. The team reports a 3.9x speedup on a single B200 and a higher UniPercept score than the official H3-Regenerate-2K API.
Pika releases H3 Flex, an open-weight video model that generates synchronized audio in six sampling steps
Pika Labs has published Pika H3 Flex, a post-trained H3 model that generates video with synchronized audio in six sampling steps and accepts up to 12 reference inputs. Open weights run locally at native 480p and 768p, while Pika's hosted API adds 1080p, 2K and 4K output.
Deep Principle releases Hydrogen-1.0, a 753B open-weight scientific model under the MIT License
Deep Principle has released Hydrogen-1.0, a 753B open-weight model for chemistry, materials science and life sciences. It is post-trained from Z.ai's GLM-5.3 and ships under the MIT License. The release reports a FrontierScience-Research win over GPT-6-Astra.
Comments
Loading...