Microsoft's GitHub now supports AI models from Anthropic and Google.

• Developers can select large language models from Anthropic and Google for GitHub Copilot Chat, in addition to models from OpenAI.
• After the launch of OpenAI's o1 models for reasoning over complex problems in September, GitHub CEO Thomas Dohmke stated in an interview that the options emerged.
• Dohmke stated in a blog post that there isn't a single model that can be applicable to all situations.

While Microsoft has a very expensive and very public relationship with OpenAI, its most successful AI product, GitHub Copilot, is now offering developers more flexibility by moving beyond OpenAI to provide a wider range of models to choose from.

Microsoft-owned GitHub announced on Tuesday that developers can use Anthropic's Claude 3.5 Sonnet model or Gemini 1.5 Pro model as alternatives to OpenAI's GPT-4o to power the GitHub Copilot Chat feature.

"GitHub CEO Thomas Dohmke stated in a post that there is no single model that can work for all scenarios, and developers expect the agency to use the models that are most effective for them. He added that the next phase of AI code generation will be characterized by multi-model functionality and choice, and that today's delivery reflects this."

In 2021, Microsoft launched GitHub Copilot, a tool that provides developers with source code suggestions. Copilot leverages models from OpenAI, which has garnered significant funding from Microsoft and experienced rapid growth following the release of ChatGPT in late 2022.

The Copilot Chat on GitHub's website and the open-source Visual Studio Code text editor will soon have OpenAI's o1-preview and o1-mini, which are designed to solve complex problems. These tools are currently available in a public preview. Google's model will also be released in public preview in the near future, according to a spokesperson.

In September, the arrival of the o1 models from OpenAI prompted GitHub to consider adding a dropdown menu for Copilot to offer more options, according to Dohmke in a CNBC interview last week. At that moment, it seemed like the ideal time to incorporate models from Anthropic and, Dohmke added.

Developers can choose the most suitable model for a specific programming language or adhere to corporate requirements, he stated.

Microsoft CEO Satya Nadella was reportedly angered by the sudden ouster and reinstatement of OpenAI CEO Sam Altman in November, which frayed the relationship between the two companies. Despite this, many other Microsoft products, including Teams and Windows, continue to use OpenAI models. Microsoft also continues to provide computing power to OpenAI, which is now valued at $157 billion.

In February, Microsoft partnered with early-stage AI model developer Mistral. The following month, Microsoft announced it was hiring former DeepMind co-founder Mustafa Suleyman and employees of his startup, Inflection, whose Pi assistant competed with OpenAI's popular ChatGPT. Over the summer, Microsoft named OpenAI as a competitor in a regulatory filing.

Dohmke stated that GitHub has partnered with engineering teams from Anthropic and Google in the past month to enhance the safety, security, and scalability of their models.

He stated that although they plan to expand the list in the future, they currently have no partnerships to announce, and GitHub aims to avoid overwhelming developers with too many options.

On Tuesday, GitHub announced a public preview of Copilot's ability to quickly review code updates through automated means.

Spark, a feature on GitHub, allows developers to create app prototypes using text input and fine-tune the designs with minimal clicks. The underlying code for the prototypes can be modified, and those interested in using Spark can sign up for the waiting list.

The GitHub spokesperson stated that once people gain access to Spark, they will have the option to use either GPT-4o or Anthropic's Claude 3.5 Sonnet model.

GitHub CEO states that the AI model is not sentient, and human developers remain in control.