Black Forest Labs Releases FLUX.1 Krea [dev], Open-Weights Image Model Trained to Avoid 'Oversaturated AI Look'
Black Forest Labs released FLUX.1 Krea [dev], an open-weights text-to-image model developed with Krea AI. The model was specifically trained to generate realistic images that avoid the oversaturated textures common in AI-generated imagery, achieving performance on par with FLUX1.1 [pro] in human preference tests.
FLUX.1 Krea [dev] Released as Open-Weights Alternative to Commercial Image Models
Black Forest Labs released FLUX.1 Krea [dev], an open-weights text-to-image generation model developed in collaboration with Krea AI. The model was specifically trained to address oversaturated textures in AI-generated images and achieve higher photorealism.
Performance and Availability
According to Black Forest Labs, FLUX.1 Krea [dev] performs on par with the closed-source FLUX1.1 [pro] in human preference assessments and outperforms previous open text-to-image models. The model is architecturally compatible with the FLUX.1 [dev] ecosystem, allowing developers to use existing tools and workflows.
Weights are available through the Black Forest Labs HuggingFace repository. Commercial licenses can be obtained through the BFL Licensing Portal. API endpoints are provided by FAL, Replicate, Runware, DataCrunch, and Together.AI.
Technical Approach
The model's training focused on generating diverse, realistic images while avoiding what the team calls the "AI look" — the oversaturated, overly processed appearance common in many text-to-image models. Black Forest Labs refers to FLUX.1 Krea [dev] as "opinionated" due to its distinctive aesthetic choices.
FLUX.1 Krea [dev] is the open-weights version of Krea 1. The architectural compatibility with FLUX.1 [dev] means the model can serve as a base for fine-tuning and customization in downstream applications.
What This Means
This release continues the trend of foundation model developers partnering with specialized AI labs to create targeted variants. By releasing open weights, Black Forest Labs enables researchers and developers to build custom versions for specific use cases while maintaining compatibility with the broader FLUX ecosystem. The focus on addressing aesthetic issues in AI-generated images suggests growing attention to the subtle quality problems that distinguish AI outputs from human-created content, though real-world performance will depend on independent testing beyond the company's own preference assessments.
Related Articles
Nvidia releases Nemotron 3.5 Super VL: open 120B multimodal MoE with 12B active parameters
Nvidia has released Nemotron 3.5 Super VL, an open multimodal Mixture-of-Experts model with 120 billion total and 12 billion active parameters, built for continuously running enterprise agents. Two compact Bolt models for local PCs and edge devices launch alongside it, though the 4 GB variant ships later this month.
Black Forest Labs releases FLUX 3 Image Klein, a 5.6B-parameter open-weights image model
FLUX 3 Image Klein is a 5.6B-parameter open-weights image model covering text-to-image, multi-reference generation and image editing. Its developer claims it leads its own blind evaluations in all three tasks. Pricing, license terms and benchmark data were not disclosed in the announcement.
OpenBMB's MiniCPM-V 4.7 35B-A3B MoE claims 13+ tokens/s on iPhone 17 Pro Max in 3 GB
OpenBMB has released MiniCPM-V 4.7 35B-A3B, the first mixture-of-experts model in the MiniCPM-V series, with open weights on Hugging Face. The company claims it runs at 13+ tokens/s using 3 GB of memory on an iPhone 17 Pro Max, and that it beats Qwen3.6 35B-A3B on overall score in less inference time in Instruct mode.
Tencent releases Hy-Memory-30B-A3B, a 3B-active model that writes agent memory for Mem0 and MemOS
Tencent Hunyuan released Hy-Memory-30B-A3B, a 30B-parameter mixture-of-experts model with about 3B active parameters, built to run the write path in external memory frameworks. Tencent reports a 55.49 cross-harness average, ahead of Claude Opus 5 (54.62) and DeepSeek V4.1 Flash (53.64) on its own fixed 20% benchmark subsamples.
Comments
Loading...