DeepSeek's V4 large language model could potentially offer enterprise clients a more cost-effective technical solution for Microsoft's AI tools.
According to a report, Microsoft is considering integrating the V4 large model from Chinese AI company DeepSeek as a lower-cost alternative technical solution for its Copilot collaboration and intelligent assistant.
The report states that the U.S. tech giant is testing a locally deployed version of DeepSeek's V4 model, along with other open-source large models, on Microsoft's cloud infrastructure. This initiative aims to replace the current Anthropic and OpenAI models that power Copilot, with the goal of reducing service costs. Microsoft plans to launch this low-cost large model service to the market within the coming weeks.
DeepSeek's open-source V4 model, released in April of this year, has gained rapid popularity among global developers and the research community due to its cost-performance advantages and is widely used in AI code development scenarios. Earlier this week, it was reported that DeepSeek had completed its first funding round, raising over $7 billion.
The report indicates that if Microsoft finalizes its decision to adopt the DeepSeek V4 model, the model would be deployed on Microsoft's Azure cloud platform, with all customer data remaining within Microsoft's cloud service system.
Comments