Google DeepMindMar 12, 2025
Gemma 3
Google's family of lightweight open models from 1B to 27B parameters, with image input on larger sizes, 128K context and support for 140+ languages.
- Type
- Small & On-device
- Context
- 128K
- Weights
- Open
Tools, Software & Models / Models Directory
Compact models for laptops, phones and the edge.
Models directory
Google's family of lightweight open models from 1B to 27B parameters, with image input on larger sizes, 128K context and support for 140+ languages.
Microsoft's 14-billion-parameter small language model trained heavily on high-quality synthetic data, strong at math and reasoning for its size.
Distillation trains a compact "student" model to imitate a large "teacher". How it works, why it powers today's small models, and its legal and quality limits.
AI Desk3 min read