AI coding assistants cost more than the plan price: Copilot, Claude Code, and Codex compared

Every AI coding assistant announces a monthly plan price, and almost none of them charge that price. In 2026 the difference comes from metering: GitHub Copilot moved to usage-based AI Credits on June 1, Claude Code is bundled into Claude Pro and Max, and Codex is bundled into ChatGPT Plus and Pro. The plan price is the entry ticket; the real cost depends on how much context your sessions push through the models.

GitHub Copilot lists four individual plans: Free at $0 with 2,000 completions and limited chat and agent usage; Pro at $10 with unlimited completions and $15 of monthly AI Credits; Pro+ at $39 with premium models such as Opus and $70 of monthly AI Credits; and Max at $100 with priority access and $200 of monthly AI Credits. AI Credits are consumed by input, output, and cached tokens at published per-model rates, so one long agent session can spend a meaningful share of the monthly allotment.

Claude Code ships inside Claude Pro at $20 a month and Max at $100 or $200, and Codex ships inside ChatGPT Plus at $20 and Pro at $100 or $200. That means the cheapest way to try either agent is often the existing consumer plan you already have, not a separate tool subscription. The practical choice is between one broad assistant with metered credits (Copilot), and agents whose capacity resets with the plan (Claude Code, Codex).

Pick by workflow, not by headline: if you live in GitHub for reviews and PRs, Copilot is the natural fit; if you want a terminal agent that can run long autonomous sessions, Claude Code or Codex inside your existing plan is the cheaper start. The limitation of this comparison is that credits and included usage change frequently, so re-check the official pricing pages before committing to a plan.