The frontiers of pixel generation
With Andreas Blattmann, co-founder of Black Forest Labs at RAAIS 2025.
At RAAIS 2025, Andreas Blattmann, co-founder of Black Forest Labs, shared a deep dive into FLUX.1 Kontext, the startup’s newly released generative model for controllable image and video generation. The talk gave a rare look under the hood of one of the most technically ambitious and commercially relevant AI-first companies to emerge in Europe.
Andreas is no stranger to the field. He was among the original researchers behind Stable Diffusion and later worked with Stability AI before striking out on his own to build the model he always wanted: a unified, fast, and open infrastructure for pixel generation, whether from text, images, or combinations of the two.
From the Black Forest to the frontier of pixels
Founded less than a year ago, Black Forest Labs is headquartered in Freiburg, Germany, with a growing team of 30. Its mission is focused: build frontier models for text-to-image and text-to-video generation. Rather than pursuing a consumer product, Black Forest Labs offers its models through APIs, designed with B2B use cases in mind.
In a notable departure from closed commercial offerings, the company is deeply committed to open source. They release model weights for non-commercial use, and licensing them for commercial applications.
FLUX.1 Kontext
The centerpiece of Andreas’ talk was FLUX.1 Kontext, the lab’s latest frontier model. At a high level, it fuses text-to-image generation and image editing into a single unified system. This differs from traditional approaches that required multiple models, custom pipelines, and fine-tuning for each new task.
Why does this matter? Today’s image generation models have mastered complex spatial composition, photorealism, and even basic text rendering. But when it comes to editing, especially iterative edits, character preservation, style transfer, or complex instruction following, most systems fall short. They’re slow, brittle, and require fine-tuned workflows for every use case.
FLUX.1 Kontext introduces a zero-shot, multi-modal interface that lets users:
Generate images from text (the classic "text-to-image" path),
Edit existing images using instruction prompts (e.g. “remove the star from her face”, “make this into a Lego scene”, “extract the skirt and put it on a white background”), and
Perform style transfer, object removal, character consistency, and typography editing all without retraining or switching models.
Fast, general, and iterative
Speed is a key differentiator. FLUX.1 Kontext generates one-megapixel images in just 2.6 seconds, over 20x faster than GPT-4 Image, with competitive output quality. For image-to-image tasks (like editing), it’s 10x faster than comparable models. That speed unlocks real-time, iterative workflows: editing product images, tweaking branding assets, or generating personalized content for marketing, all in seconds.
It’s also available to try. Black Forest Labs has launched a public playground accessible at playground.bfl.ai where anyone can interact with FLUX.1 Kontext in a no-code interface.
A foundation for the visual internet
Andreas’ final call was pragmatic and pointed: “Everyone needs marketing images, everyone needs visual content.” In a world where pixels are becoming programmable, tools like FLUX.1 Kontext are infrastructure, not novelty.
With a performant, open, and composable foundation, Black Forest Labs is poised to be the go-to stack for image and video generation in B2B workflows from e-commerce to entertainment.
