Visual Studio 2022 adds Copilot memory and Anthropic models

Microsoft’s October 2025 update reshapes how developers use Visual Studio 2022 by adding persistent memory to GitHub Copilot, widening the chat model choices, and setting out a C++ migration path toward Visual Studio 2026.

With GitHub Copilot Memories, the assistant can now retain corrections, declared standards, and explicit prompts to remember, then offer to save those preferences into .editorconfig, CONTRIBUTING.md, or README.md so the rules travel with the project and apply to the entire team. Storing preferences in code and documentation files shifts AI guidance from ephemeral chat to visible governance.

Developers also gain new model options in chat with access to Anthropic’s Claude Sonnet 4.5, described as stable and versatile for software development, and Claude Haiku 4.5, which emphasizes higher performance at lower cost, while Microsoft notes additional AI features on its developer blog and positions GitHub Copilot as a neutral entry point for multiple LLM providers.

To prepare for the next major release, Microsoft has published a guide that walks C++ teams through migrating to Visual Studio 2026, available today in the Insiders program and installable alongside a stable version so existing MSVC toolsets can remain in place until the setup wizard adds any missing components and the MSCV Build Tools are finally upgraded to version 14.50 with the MSVC compiler at 19.50, a pace that gives space to evaluate GitHub Copilot without disrupting toolchains.

The timeline also intersects with the community’s calendar as the online conference betterCode() .NET 10.0 on November 18, 2025 gathers .NET professionals to discuss improved classes in .NET 10.0 and Native AOT with Entity Framework Core 10.0, followed by six full day workshops on C# 14.0, artificial intelligence, and web APIs, which together outline where GitHub Copilot will factor into daily development practice.

In the broader software market, the addition of persistent repository memories and a multi model chat inside the IDE points to a maturing stage where buyers expect flexibility rather than vendor lock in, and where GitHub Copilot competes on workflow depth as much as on raw model capability.

For startups, including those in the UK’s crowded software scene, integrating memory into the development environment reduces onboarding friction and codifies standards from day one, a posture that can help teams justify investments in platform tooling as GitHub Copilot becomes a common baseline across repositories.

The direction is clear as Microsoft ties IDE infrastructure to model choice and codified rules, and the signal for enterprise software leaders is that GitHub Copilot is moving from a helpful plugin to a governed system for producing compliant, reusable project content.