Read the price page before you connect a repository
Five steps for reading a coding tool's own plan page, and separating a chat subscription from a bill for API use, before the tool can see a repo.
Updated 2026-11-09 · Sources checked 2026-11-09 · The Loaded Tokens desk
GitHub publishes Copilot plan names and prices on its plans page. Cursor publishes plan prices on its models and pricing page. Anthropic publishes Claude prices on its pricing page, and Claude Code's product overview is a separate document. Start there. A screenshot in a chat thread is not the price.
GitHub's plans page lists Copilot Free at $0, Pro at $10 per user each month, and Pro+ at $39 per user each month, with plan credits and model restrictions. Cursor's models and pricing page lists Free at $0, Plus at $8 a month, and Premium at $30 a month, and says third-party models on eligible tiers use prepaid credit at API list price. We copied those figures onto the matching tool pages on November 7, 2026. If you are buying today, read the vendor page again.
Claude Code is described in its official overview as an agentic coding tool for repository and terminal workflows. Our Claude Code profile, citing Anthropic's pricing page, says it is included with eligible Claude plans, including Pro, and that API-backed usage follows separate billing. A plan that covers the chat box does not automatically cover every command the tool runs.
The Claude Code overview documents repository and terminal workflows. A terminal that can run commands is a different risk from a completion inside one open file. Cursor's product page describes an AI coding environment with agent workflows. Write down whether the next step can edit files, run commands, or only suggest a line. Stop if the docs do not say.
Choose a small change you can read, with tests, and without deployment rights. That is the evaluation described on our Claude Code page. It is a suggested check, not a test Loaded Tokens ran, and it is not a promise about speed or correctness.
Questions
Is this a ranking of coding agents?
No. It is a way to read three public product and pricing pages before a tool can see a repository. The verdicts live on the tool pages, and they are not a race result.
Did Loaded Tokens run these tools for this guide?
No. The steps use the vendor pages linked below and the dates already printed on our tool profiles. We are not reporting a hands-on test.
Sources
Every product fact above comes from one of these pages. We did not run a separate test for this guide.