Microsoft has officially launched two new compact language models: Phi-4-multimodal and Phi-4-mini. Starting today, these models are available simultaneously on the Azure AI Foundry and Hugging Face platforms.

Phi-4-multimodal has demonstrated impressive capabilities across various domains, including enhanced speech recognition, translation, text summarization, audio comprehension, and image analysis. This advancement signifies that the model can deliver more accurate and efficient performance when handling complex multimodal data.

On the other hand, Phi-4-mini focuses on optimizing speed and efficiency, making it ideal for scenarios with strict computational resource constraints. Whether on smartphones, personal computers, or in-car systems, developers can flexibly utilize this model to meet diverse needs.

The release of these two models further expands Microsoft's presence in the artificial intelligence field, offering users and developers a broader range of options.