Blog
Mert Özen's blog posts on software development, backend, frontend, and DevOps.
Copilot in Teams and Organizations: Policies, Content Exclusion, Model Access, Usage Metrics, Cost and a Comparison with Claude Code/Codex
In an organization, Copilot's cost is driven mostly by the model choice and code review on every push: for an eight-person Business team, Sonnet sessions plus a Balanced review on every push use up the shared pool in a ninth of the month, with two thirds of the credits going to review. Policy layers, content exclusion, managed settings and the MCP allowlist, metrics, the audit log, budgets and a comparison with Claude Code and Codex.
Getting Started with GitHub Copilot: Plans, AI Credits, Surfaces (VS 2026, VS Code, Rider, GitHub.com, CLI) and Model Choice
In Copilot, agent sessions spend your credits, not inline suggestions: since 1 June 2026 billing uses token-based GitHub AI Credits, and a thirty-step session costs between 4 and 128 credits depending on the model. Plans, base and flex credits, auto and Copilot Fusion, setting up Visual Studio 2026, VS Code, GitHub.com and the CLI, and organization policies.