Launched preview of flagship AI model HY3 with 295B parameters.

Open-sourced MegaStyle, a 1.4M AI-generated image dataset for style transfer

Open-sourced HY-World 2.0, a multimodal 3D world model for generation, reconstruction, and simulation

Open-sourced HY-World 2.0, a 3D world model that generates navigable environments in a single forward pass.

Launched 'Kernels' hub on its platform for sharing and discovering optimized GPU code.

Announced completion of a large-scale project converting 27,000 arXiv papers from PDF to Markdown using an open 5B model.

Launched the Tencent Advertising Algorithm Challenge 2025 and released the TencentGR-1M and TencentGR-10M datasets.

Julien Chaumond, CTO of Hugging Face, announced the Safetensors transfer via social media.

Massive open-source dataset of 48,478 computer screen recordings released on Hugging Face

Launched Penguin-VL vision-language model with LLM-initialized vision encoder