Recent developments in AI model training have introduced new capabilities for users of FLUX LoRAs. Users can now train FLUX LoRAs on the FAL platform and directly upload them to Hugging Face. Additionally, a new Replicate model has been launched that integrates XLab-AI’s v3 controlnets, enabling features such as preprocessor selection, image-to-image processing, and control strength adjustments. Furthermore, a Google Colab Notebook named RunFlux has been released, allowing users to initiate FLUX1-dev LoRA training by uploading images and captions, selecting a GPU, and monitoring the training process, with results pushed to their Hugging Face repositories. The integration of FLUX LoRAs into the controlnet model has also been confirmed, supporting platforms like Replicate, Hugging Face, and Civit.