Google's Gemini 3.1 Pro now available in GitHub Copilot public preview

Gemini 3.1 Pro Reaches GitHub Copilot Users

Google has rolled out Gemini 3.1 Pro to GitHub Copilot in public preview, marking the first wide availability of the company's agentic coding model for software developers. The deployment represents a significant expansion of Gemini's presence in Microsoft's development tools ecosystem.

Model Capabilities

According to GitHub's announcement, Gemini 3.1 Pro excels in executing effective and efficient edit-then-test loops—a critical workflow pattern in modern development where the model iterates on code changes and validates them through testing. The model is designed with agentic capabilities, enabling it to take actions, use tools, and maintain context across multiple development tasks.

Google positions this as a model specifically optimized for coding workflows, distinct from general-purpose language models. Early testing has focused on the model's performance in autonomous coding scenarios where multiple iterations and tool interactions are required.

Public Preview Status

The public preview indicates this is not yet a stable release. GitHub users can begin testing Gemini 3.1 Pro's performance on their codebases, with feedback likely to inform subsequent iterations before a general availability release. GitHub Copilot subscribers with access to preview features will gain early exposure to the model.

This preview deployment also signals Google's strategy to compete directly in the AI-assisted coding space, where Claude (Anthropic) and GPT-4 (OpenAI) currently dominate. By integrating Gemini 3.1 Pro into Copilot—Microsoft's widely adopted coding assistant—Google extends its reach to millions of developers.

What This Means

The Gemini 3.1 Pro preview in Copilot represents meaningful competition in enterprise coding AI. While specific benchmark scores and performance metrics were not disclosed in the announcement, the emphasis on edit-then-test efficiency and agentic behavior suggests Google is targeting real-world development workflows rather than benchmark optimization. For GitHub Copilot users, this adds another model option and potentially improves overall performance quality through competition. The timing also indicates Google is actively pushing Gemini as a viable alternative to established coding models, though adoption will depend on actual performance data that will emerge from public preview testing.