GitHub Copilot goes ‘multi-model’ with support for four new LLMs

Builders, rejoice: You will quickly be capable to use a number of new massive language fashions (LLMs) in GitHub Copilot, the corporate’s coding assistant.

On Tuesday at its annual GitHub Universe convention, the Microsoft subsidiary introduced help for 4 new LLMs in Copilot: Claude 3.5 Sonnet, Gemini 1.5 Professional, and OpenAI’s o1-preview alongside o1-mini. The OpenAI fashions can be found now in Copilot Chat, with Claude 3.5 Sonnet up subsequent, adopted by Gemini 1.5 Professional “within the coming weeks,” in line with the announcement.

“From Copilot Workspace to multi-file enhancing to code evaluate, safety autofix, and the CLI, we are going to convey multi-model alternative throughout a lot of GitHub Copilot’s floor areas and capabilities quickly,” the corporate famous. 

GitHub first launched Copilot with Codex, a nascent model of OpenAI’s GPT-3. Final 12 months, GitHub launched Copilot Chat, first with GPT-3.5 after which later GPT-4. The corporate says it has frequently up to date its base fashions relying on high quality and latency wants with every part from GPT-3.5 Turbo to GPT-4o mini.

GitHub says it is seen a “growth” within the capability of each small and enormous LMs to serve totally different programming wants. 

“The subsequent part of AI code technology is not going to solely be outlined by multi-model performance, however by multi-model alternative,” GitHub stated within the announcement. “GitHub is dedicated to its ethos as an open developer platform, and guaranteeing each developer has the company to construct with the fashions that work greatest for them.”

The corporate additionally launched GitHub Spark, an AI device for constructing apps fully in pure language. With it, customers can create Sparks, or “micro apps” that leverage AI and exterior information with out consuming up cloud area. Join an early preview at GitHub Spark.