Microsoft on Tuesday announced the release of Phi-3-mini, a new, more affordable artificial intelligence (AI) model. This lightweight model is designed to provide similar functionality to previous models at a fraction of the cost. Sébastien Bubeck, vice president of GenAI Research at Microsoft, highlighted the economic benefits of the new model. “The Phi-3 is not just slightly cheaper, it's dramatically cheaper. We're looking at about a 10x cost reduction compared to other models on the market with comparable features,” Bubeck said. Reuters reported that he said this at the time of the announcement.
What can you do with Phi-3-mini?
Phi-3-mini is part of Microsoft's broader effort to introduce a set of Small Language Models (SLMs) that are tailored for simpler tasks and are ideal for businesses with fewer resources.
This approach is part of the tech giant's strategic bet on technology that it believes will have a major impact on global industry and workplace dynamics.
Microsoft claimed that the Phi-3-mini is measured with 3.8 billion parameters and can outperform models twice its size.
Spotlight: Sora AI video tools are already being leveraged by filmmakers to create short films.The results are “surreal”
How is an LLM different from an SLM?
One of the key features of the SLM-based Phi-3-mini is its compact size, which allows it to operate efficiently on a local machine without requiring extensive processing power.
Unlike large-scale language models (LLMs) that require multiple parallel processing units, Phi-3-mini can quickly generate data with minimal hardware, making it accessible to small businesses and individual users. It will look like this.
Where can I get Phi-3-mini?
Microsoft is making Phi-3-mini available on several platforms, including its own Azure AI Model Catalog, Hugging Face, a popular machine learning model platform, and Ollama, a framework that makes it easy to run models on your local machine. We made it available immediately. Additionally, this model is optimized for his Nvidia's GPUs and supported by Nvidia Inference Microservices (NIM), which improves performance and versatility.
The launch follows Microsoft's recent $1.5 billion investment in UAE-based AI company G42 and reflects the company's continued efforts to expand its AI footprint globally. . The company also partnered with French startup Mistral AI to integrate its model into the Azure cloud computing platform, strengthening Microsoft's position as a leader in AI innovation and accessibility.
