Microsoft has launched Phi-3-mini, a highly compact artificial intelligence system, that aspires to provide affordable solutions in the AI world with its fast-changing trend.
This is the first of three SLMs from Microsoft. The company’s strategy to meet the demands of a broader crowd and transform how people engage with technology at different levels of society is to be commended.
Sébastien Bubeck, Microsoft’s VP of GenAI research, indicated that the Phi-3-mini‘s price is much more reasonable than that of its competitors available on the market. He highlighted that the savings are not just as an order but rather as an order of magnitude which can go up to ten times cheaper than those with similar specifications.
Phi-3 Mini boasts a parameter count of 3.8 billion and is trained on a dataset that, while smaller compared to large language models like GPT-4, still ensures robust performance and versatility.
Phi-3-mini, designed for less complex tasks, delivers practical solutions that are especially suited to businesses with limitations. This strategic direction highlights that Microsoft will make AI democratic and readily available to companies of all sizes and resourcing.
Together with the increased availability of Phi-3-mini via Azure AI, the Hugging Face model platform, and Ollama, the local machine learning model deployment platform, the model has been a major stride towards the goal of shifting human tasks toward the AI co-worker.Â
Moreover, the SLM is GPU optimized and is natively integrated with Nvidia microservices which is known by the name Nvidia Inference Microservices, enabling easy deployment of a pair device with the best accessibility and performance.
One of the most recent actions of Microsoft is investing $1.5 billion into G42, AI AI-based company based in the United Arab Emirates. On the contrary, Microsoft has developed strategic partnerships with cutting-edge startups like Mistral AI and facilitated the easy integration of the latest AI models in the Azure Cloud platform.
