Microsoft Research has released Phi-2, a small language model with 2.7 billion parameters.

Phi-2 outperforms larger models in various benchmarks, and matches or outperforms Google's Gemini Nano 2 model.

The model's success is due to high-quality training data and innovative scaling techniques.

Phi-2 can run on low-tier equipment and is available in the Azure AI Studio model catalog, promoting the democratization of AI research.

The model prioritizes safety and surpasses open-source models in terms of toxicity and bias.

Microsoft's Phi-2 and Orca 2 models demonstrate the company's commitment to powerful AI technologies and not solely relying on OpenAI.