Microsoft is moving to a multi-model approach with GitHub Copilot. This means the tool is no longer working exclusively by OpenAI’s GPT models.
GitHub Copilot will soon allow developers to choose whether the code assistant should run on a GPT model. Then, developers will also be able to request assistance from Claude and Gemini.
Mutual competition
In the coming weeks, Anthropic’s Claude 3.5 Sonnet and Google’s Gemini 1.5 Pro will be available in Copilot Chat. The chat feature serves just as a starting point for integrating new models in GitHub Copilot, and the rollout will be extended to various parts of Copilot, including Workspace.
Tip! GitHub Copilot Workspace: will AI replace software engineers?
The competition between the different models is fierce. Claude 3.5 Sonnet did show in the benchmarks that it was better at programming than GPT-4o. On the other hand, Microsoft is also expanding the range of OpenAI models with the arrival of o1-preview and o1-mini.
Shift to multi-model selection
Since its launch, the tool has been tied to OpenAI’s LLMs because of the close relationship between OpenAI and Microsoft, the maintainer of GitHub. Microsoft is now working its way out of the alliance, although this is only a tiny beginning since GitHub has continued to operate largely independently despite its acquisition by Microsoft.
According to GitHub CEO Thomas Dohmke, the multi-model approach is simply a logical consequence of the developments happening in the AI world. “The next phase of AI code generation will not only be defined by multi-model functionality, but also by multi-model selection,” he writes in the blog.
Also read: OpenAI launches o1, making ChatGPT smarter than ever