Apple introduces new AI models ahead of September 9 event
A week before Apple's "Awe Dropping" event on September 9, the company has rolled out two fresh AI models: FastVLM and MobileCLIP2.
Both are built to run directly on your device for super-fast results—think instant image analysis and image captions.
FastVLM handles high-res visuals in a snap, while MobileCLIP2 blends vision and language skills for tasks like object recognition and captioning.
Both models are available on Hugging Face
You can find both models on Hugging Face, optimized especially for Apple silicon using their open-source tech.
MobileCLIP2 stands out by being 85 times faster and way smaller than older versions, making it perfect for things like object recognition or generating quick image descriptions.
Plus, there's a browser version of FastVLM (called FastVLM-0.5B) that lets you summarize images or spot text right from your browser—no downloads needed.
Apple's keeping user privacy front and center with these updates.