Microsoft is pushing to diversify artificial intelligence by driving its flagship AI product, Microsoft 365 Copilot, by adding more in-house and third-party models, sources told Reuters. The move will help to reduce costs and lessen dependence on OpenAI, the AI startup that has been central to Microsoft’s AI strategy.
Read More: OpenAI Rolls Out ChatGPT’s Search Feature for All Users
Since its launch in March 2023, 365 Copilot has relied heavily on OpenAI’s GPT-4 model. However, Microsoft has been training smaller, in-house models, including Phi-4 and customizing open-weight models for affordability. “OpenAI remains our collaborator on frontier models,” said a Microsoft spokesperson, emphasising the ongoing partnership alongside diversification efforts.
Other business units, such as GitHub, have followed a similar approach. In October, GitHub began using AI models from Anthropic and Google in addition to OpenAI’s GPT-4. Likewise, the consumer chatbot Copilot now employs a mix of in-house and OpenAI technologies.
Despite concerns about pricing and utility, 365 Copilot shows promising adoption rates. BNP Paribas Exane analysts predict over 10 million paid users by year-end. Microsoft reports that 70% of Fortune 500 companies have already embraced the tool. By integrating a diverse range of AI models, Microsoft aims to balance costs while enhancing the product’s value for users.
Stay tuned to Brandsynario for latest news and updates.