GitHub Copilot AI Credits Billing Explained: What's Free, What's Metered, and My Hybrid Claude Code Workflow

Short summary

GitHub Copilot switched to usage-based AI Credits billing on June 1, sparking backlash over unpredictability rather than price—code completions stay unlimited while chat, agents, and PR review now consume credits. Claude Code and Cursor gained significant traction, with developers averaging 2.3 AI tools simultaneously, prompting the author's hybrid workflow centered on AGENTS.md.

• •Copilot credits model: code completions unlimited, chat/agents/PR review metered at 1 credit = 1¢
• •Community anger stems from unpredictability and deprecation of GPT-4.1, not the price itself
• •Claude Code reached 91% CSAT and 54 NPS; workplace adoption tied at 18% with Cursor, up 6x since mid-2025

Generated with AI, which can make mistakes.