Phi-3: Microsoft relaunches SLMs with quality of 10x bigger models
In one sentence Microsoft releases Phi-3-mini 3.8B, small 7B, medium 14B. Mini runs on iPhone and beats Mixtral 8x7B on many benchmarks. Confirms the 'curated data > scale' thesis.
While everyone races to make bigger models, Microsoft Research goes the other way: small models trained on data curated like a textbook. Phi-3-mini has 3.8 billion parameters (much less than Mixtral 8x7B) yet rivals far larger models.
The practical bit: the model runs on an iPhone 14, offline, without internet. Microsoft shows a demo where an iPhone answers complex questions without any connection.
For developers it means useful AI can fit inside a mobile app, a router, an industrial PLC, without paid API calls or GPU ops. For privacy it is huge: data never leaves the device.
Companies
Microsoft
Tools
Phi-3-mini, Phi-3-small, Phi-3-medium
Tags
Sources