Microsoft has released a new family of AI models called Phi-3.5. These models are lightweight and built on synthetic data and filtered public websites. All models are now available on Hugging Face.
Phi-3 RSS
Google expanded its Gemma 2 family of language models with the new 2 billion parameter Gemma 2 2B. It outperforms GPT-3.5 models and can run efficiently on various devices including PCs.
During Build 2024, Microsoft announced another member of the Phi AI small language model family, called Phi Silica. It will be able to be run locally on the NPU chips in Copilot+ PCs.
During Build 2024, Microsoft announced a preview version of Phi-3-vision. This is a multimodel addition to its Phi-3 AI small language model family that was first revealed in April.
Microsoft has introduced an upgraded lineup of small AI models, Phi-3. The family includes Phi-3 Mini with 3.8 billion parameters, Phi-3 Small with 7 billion, and Phi-3 Medium with 14 billion.