Hugging Face推出的SmolVLM多模態模型迎來重大突破:通過WebGPU技術，SmolVLM現可在瀏覽器中實現實時網絡攝像頭圖像識別，無需服務器支持，全部計算在用戶設備上完成。這一創新

A major breakthrough has been achieved by SmolVLM, a multimodal model introduced by Hugging Face:

Salesforce AI Research在Hugging Face平臺正式發佈BLIP3-o應用，這款全開源的統一多模態模型家族以其卓越的圖像理解與生成能力引發業界熱議。BLIP3-o通過創新

Salesforce AI Research在Hugging Face平台正式发布BLIP3-o应用，这款全开源的统一多模态模型家族以其卓越的图像理解与生成能力引发业界热议。BLIP3-o通过创新

Salesforce AI Research has officially released BLIP3-o on the Hugging Face platform, a fully open-