Microsoft just launched its latest language model that specialises in complex reasoning called Phi-4.
In a blog post released earlier today, Bill Gates-founded the organisation revealed its latest generative artificial intelligence (AI) product, the Phi-4 which is a small language model (SLM) in the Phi series.
Phi-4 is a 14 billion-parameter small language model (SLM) that excels in complex reasoning, particularly in areas like mathematics and traditional language processing tasks.
Microsoft says that the new Phi-4 is currently available on Azure AI Foundry under a Microsoft Research License Agreement (MSRLA) and will be available on Hugging Face next week.
This AI-driven language model is especially great at solving mathematics problems. Microsoft said that Phi-4 outperformed much larger models, including Gemini Pro 1.5, on math competition problems.
Phi-4 was trained on a combination of high-quality synthetic datasets and high-quality human-generated content along with other post-training improvements.
Also Read: Microsoft Releases New AI Capabilities to Copilot Studio & Dynamics 365
How to use Microsoft Phi-4?
Microsoft Phi-4 is a new-gen AI that can be accessed by software developers or organisations. While currently available primarily through Azure AI Foundry under a research license agreement, it is expected to be released on Hugging Face in the future.
This will allow developers to optimise Phi-4’s advanced reasoning capabilities in solving mathematical problems especially. Once the necessary permissions are sought, individuals can use Phi-4 for their research projects.
Also Read: Microsoft Launches Healthcare AI Tools to Aid Doctors & Nurses
Is it safe?
According to Microsoft, Phi-4’s safety is of the utmost importance. The tech giant on several occasions has emphasised responsible AI development.
This is why the gen-AI model’s accessibility is restricted to just researchers on Azure AI Foundry as of now. The access is only granted if the researcher’s application for a research license agreement is approved.
The controlled roll-out of Phi-4 also allows Microsoft to potentially introduce a comprehensive set of safety features and monitoring tools. This would also address concerns about AI risk management.
Developers using Phi-4 can access evaluation tools to evaluate model quality and safety, as well as content filtering capabilities to prevent misuse.
Also Read: What is Microsoft Copilot Studio? How To Use it and Use Cases
