This update covers several developments in the AI landscape, including Meta opening its Muse platform to third-party developers, releasing a new version of its segmentation model, and a security incident involving Google's Gemini.
Alibaba's Qwen team has released Qwen-Image-2.1, a unified text-to-image generation and image editing model that consolidates previous separate checkpoints into a single 7B parameter diffusion transformer.
Qwen has released the open-weight Qwen-Image-2.1, a unified model designed for both image generation and editing.
Badaramoni has released a new adapter for the Krea-2-Turbo model that combines public realism packs of different formats into a single file. The release addresses the limitation where standard merger tools fail to combine classic LoRA and LoKR adapters because they do not share the same low-rank format.
Diffusers-workflow is a declarative engine built on top of Hugging Face Diffusers that allows large language model agents to author, validate, run, and inspect image or video pipelines using JSON documents rather than Python scripts. It exposes an MCP server with approximately 50 tools as the primary interface, enabling unattended agent control over GPU resources.
Gradio has released Workflow1111, a project that rebuilds most of AUTOMATIC1111's stable-diffusion-webui feature set using the gr.Workflow framework. The application consists of a single canvas containing eleven media pipelines built from seventy-three nodes, covering text-to-image, image-to-video, and various preprocessing tasks.
In collaboration with filmmaker Liz Garbus and Primordial Soup, Google DeepMind utilized generative AI models to reconstruct an unphotographed meeting between Burt and Ethelle Shatz for the documentary short "Love, Rendered." The project addresses cognitive decline by recreating a specific memory that existed only in the couple's minds.
A user has published code and weights for running NVIDIA's 64-billion parameter Cosmos3 model locally using INT4 quantization. The implementation supports both Text-to-Image and Image-to-Video tasks on Apple Silicon via MLX as well as CUDA.
OpenAI has released ChatGPT Images 2.5, an update to its image generation models that enhances instruction-following capabilities across multiple turns and increases response speed. The new version is also better at preserving subjects from reference photos provided by users.
OpenAI has released ChatGPT Images 2.5, a new state-of-the-art image model that delivers sharper details, more precise editing capabilities, and up to 50% faster generation compared to the previous version.
An engineer has open-sourced a reproduction of Surya Narreddi’s viral project, which trains a coding model to paint watercolours using JavaScript and reinforcement learning. The implementation utilizes the TRL library and OpenEnv to create an end-to-end pipeline on Hugging Face for training, scoring, and inference.
Google is rolling out Google Pics, an image creation and editing tool built on the Nano Banana model, to all Google AI Pro and Ultra subscribers and most Workspace business customers over the coming weeks.