Black Forest Labs upgraded its FLUX Video Upscaler, which can upscale video to 4K.
BFL says the upgraded version costs less than Topaz Astra 2 and runs twice as fast.
The lab behind FLUX — frontier open and commercial models for image, and now video, audio, and action prediction in one multi-modal architecture.
Black Forest Labs is the research lab behind FLUX, one of the leading families of image generation models, available in open and commercial tiers and through an API. Its latest release, FLUX 3, is a single multi-modal model spanning image, video, audio, and action prediction, jointly trained in one unified architecture, with FLUX 3 Video in early access. The unified design can even be extended to predict actions for robotics.
Pricing model: Freemium
Common questions about Black Forest Labs
Black Forest Labs upgraded its FLUX Video Upscaler, which can upscale video to 4K.
BFL says the upgraded version costs less than Topaz Astra 2 and runs twice as fast.
Black Forest Labs and fal Research have introduced FLUX 3 Max, a video model available on fal. They say it generates a 10-second 720/768p video in about 12 seconds.
fal says the model is tied for 1 in image-to-video quality on independent third-party benchmarks. Its launch offer was 75% off on fal for three weeks starting October 19, 2026.
Black Forest Labs announced FLUX 3 Image Klein, a 5.6B parameter model with open weights for text-to-image, multi-reference generation and image editing.
The lab says it leads its own blind evaluations across those tasks.
Original source
Black Forest Labs has released FLUX 3 Action, a 7B parameter World Action Model with open weights. The lab says it takes first place on the RoboLab benchmark.
Read the original announcement →
Black Forest Labs' FLUX Upscale is available as a standalone FLUX Tool and an endpoint. It regenerates existing video at higher resolutions, with 2K and 4K output. Black Forest Labs describes the output as up to native 4K.
Read the original announcement →
FLUX 3 Video is in early access through Black Forest Labs.
The lab describes FLUX 3 as one model jointly trained for image, video, audio, and action prediction in a single architecture. It says the model can also be extended to predict actions for robotics.
Official videos and tutorials for Black Forest Labs.
Other products to explore and follow.

An all-in-one AI creative platform (formerly Freepik) — AI image & video generation, the original Magnific upscaler, and a 250M+ asset library.
Create AI-generated images and cinematic videos from text prompts — 100+ curated styles, powered by state-of-the-art models.

Video editor where a team and agents cut footage with Claude Code and Codex, then review by link.
Desktop app that turns a prompt into an editable motion graphic — describe the scene, refine it on a real timeline, export video, transparent WebM, or HTML.

Applied biological computing company. mischief 1 is its image-to-video model, live on OpenRouter and fal from $0.005 a second.

The real-time everything app — posts, news, and conversation, plus the X API and a hosted MCP server for connecting AI agents to real-time data.