Live
Local Artifacts Support Lets Workers Test Repo Workflows Before DeployCo‑Packaged Optics Cut Power Use and Latency in Lambda’s 10k‑GPU AI ClusterScaling AI Agents on Kubernetes: New CNCF Projects and Operational ImplicationsGitHub Copilot model deprecation forces updates to AI‑assisted development pipelinesGitHub Copilot expands Enterprise Server access and adds new AI models – practical impact for engineersArchitecting Agentic Customer Engagement with Amazon Bedrock AgentCoreBuilding Multi‑Tenant Memory‑Aware Agents with Bedrock Agentic Retrieve and Managed Knowledge BaseBulk‑Start AI Scan for Up to 20 GitHub Repos: What Engineers Need to KnowLocal Artifacts Support Lets Workers Test Repo Workflows Before DeployCo‑Packaged Optics Cut Power Use and Latency in Lambda’s 10k‑GPU AI ClusterScaling AI Agents on Kubernetes: New CNCF Projects and Operational ImplicationsGitHub Copilot model deprecation forces updates to AI‑assisted development pipelinesGitHub Copilot expands Enterprise Server access and adds new AI models – practical impact for engineersArchitecting Agentic Customer Engagement with Amazon Bedrock AgentCoreBuilding Multi‑Tenant Memory‑Aware Agents with Bedrock Agentic Retrieve and Managed Knowledge BaseBulk‑Start AI Scan for Up to 20 GitHub Repos: What Engineers Need to Know
GitHub

GitHub Copilot expands Enterprise Server access and adds new AI models – practical impact for engineers

AI SummaryPowered by AI

GitHub Copilot’s November release adds Enterprise Server repository access in the mobile app, introduces Claude Fable 5.2 and Gemini 4 Carbon models, and expands VS Code agent features such as model comparison and inbox mode. These changes affect how AI, cloud, DevOps, and security teams provision models, manage session authentication, and plan for the upcoming migration to the Copilot harness.

The latest Copilot update brings three distinct shifts: mobile‑app support for GitHub Enterprise Server (GHES) repositories, two new AI models (Claude Fable 5.2 and Gemini 4 Carbon), and a set of experimental VS Code agent capabilities. For engineers responsible for building or operating AI‑assisted workflows, each shift alters the way resources are provisioned, authenticated, and observed.

Enterprise Server support in the Copilot app

Copilot now lets users browse and act on GHES repositories directly from the mobile app, a feature marked generally available. Access still requires a GitHub.com or GHE.com account, and cloud‑based session features remain unavailable for pure GHES instances. The practical implication is a split authentication surface: teams must maintain a cloud identity for app usage while preserving on‑prem GHES credentials for repository access. Automation scripts that previously relied on a single token may need to incorporate dual‑token handling or gateway services to bridge the two identity domains.

New AI models and data‑retention policy

Both Claude Fable 5.2 and Gemini 4 Carbon are offered to Copilot Pro+, Max, Business, and Enterprise tiers. Claude Fable 5.2 enables data retention by default, but Business and Enterprise administrators must explicitly enable its policy, which is off by default. This introduces a compliance consideration: organizations must audit whether retained data aligns with internal policies or regulatory requirements before activating the model. The models’ availability across tiers also means that CI/CD pipelines can now target higher‑capability LLMs without changing subscription levels, but they must respect the optional retention flag.

VS Code agent enhancements

The VS Code 1.146 release adds several experimental agent tools. Compare Models lets an agent locate available models, run side‑by‑side comparisons, and summarize results. Inbox mode aggregates agent sessions across workspaces, surfacing those that need attention first and allowing dismiss, hold, or snooze actions. Chat windows can be detached from the Agents pane, enabling custom layout with editors and terminals. Extension management now occurs within the Agents window, with warnings for unsupported features. Because these features are experimental, teams should treat them as optional pilots and monitor stability before embedding them in production pipelines.

Operational tweaks and accessibility

Project automation can now be scheduled to the minute or via custom cron expressions (e.g., weekdays at 9:30). Multilingual dictation supports French, Spanish, and additional languages after downloading the voice model. Issue and pull‑request filters now include custom fields such as a Project’s Size, improving query precision. Accessibility settings combine high‑contrast filters with color choices and preview them before application; a new GitHub Dimmed theme offers a dark‑only palette. These UI enhancements can reduce friction for globally distributed teams and support accessibility compliance audits.

Related CloudNinjas coverage: AI engineering.

What This Means For Practitioners

Evaluate whether your organization needs GHES integration in the Copilot app and adjust identity management to accommodate dual accounts. Review the default data‑retention setting for Claude Fable 5.2 and align it with compliance frameworks before enabling the model. Pilot the experimental VS Code agent features in a sandbox environment, and decide if Inbox mode or Compare Models adds measurable efficiency. Finally, begin migrating any remaining workflows to the Copilot harness ahead of the February 2027 deprecation of the local harness to avoid future disruption.

Originally published atGitHub Changelog