Claude Code vs Copilot (October 2026): Pricing, Claude Models and Use in China
On a budget and coding mostly in the editor? GitHub Copilot has a free plan, starts at $10 and includes Claude models. Handing off long tasks? Claude Code starts at $20. Compares billing, whether Claude in Copilot replaces Claude Code, use in mainland China and a $30-a-month split.
If your budget is tight and you mostly write code in your editor, choose GitHub Copilot: it has a free plan, paid plans start at $10 a month, and Claude models are available inside it. If you regularly hand off whole tasks and want an agent to change files across a project and keep running tests on its own, choose Claude Code, which starts with Claude Pro at $20 a month. For developers in mainland China the gap is larger still. Claude Code is outside Anthropic's supported regions, while GitHub's trade controls page does not list mainland China among the places Copilot is restricted.
An AI coding assistant built into editors and GitHub, with an agent mode
Anthropic's coding agent, run from the terminal, with IDE extensions
Free plan
Yes, 2,000 completions and 50 chat requests a month
No
Paid plans
Pro $10/month, Pro+ $39, Max $100
Pro $20/month, Max from $100
How usage is counted
Completions unlimited on paid plans; chat and agent work use AI Credits
One pool shared with Claude chat, with five-hour and weekly limits
Models
Claude, GPT, Gemini, Grok, Kimi and others
Anthropic's Claude models
Cloud coding agent
Pro and above
Web version at claude.ai/code
Mainland China
Not among Copilot's restricted destinations in GitHub's trade controls
Not an Anthropic-supported region
How Claude Code and Copilot pricing works
Copilot costs half as much and has a free plan. Copilot Free gives you 2,000 code completions and 50 chat requests a month, without the cloud coding agent. Pro costs $10 a month with unlimited completions plus $15 of AI Credits ($10 base and $5 flex). Pro+ costs $39 with $70 of credits, and Max costs $100 with $200. One credit equals $0.01, and you can buy more when you run out.
Claude Code comes with a Claude subscription. Claude Pro costs $20 a month, or $17 a month billed annually. Max starts at $100 a month in 5x and 20x Pro sizes. The allowance is shared with your Claude chats, and paid plans add weekly limits on top of five-hour session limits.
Because the two bill differently, sticker prices only tell part of the story. Completions on paid Copilot plans do not use credits, so if you code every day and lean mostly on tab completion, Pro at $10 is very good value. If you run long agent tasks, though, AI Credits drain faster and you may need to top up. A Claude subscription includes a substantial amount of agent usage in the flat price, which pays off the more long tasks you run.
Copilot already has Claude. Do you still need Claude Code?
The Copilot plans page lists several Claude models, including the Sonnet and Opus families, on every plan. That leads many people to ask whether a separate Claude Code subscription is worth it.
It depends on whether you want the model or the tool.
If you want Claude models for completion, chat and a few edits inside your editor, Copilot is enough, and cheaper. Copilot's agent mode can also research a repository, propose a plan, edit files and prepare a pull request for you to review.
If you want an agent designed for long tasks, Claude Code fits better. It runs locally in your terminal and talks directly to model APIs, and it can keep working for a long stretch on cross-directory refactors, dependency upgrades or issue-to-passing-tests work. On Pro, Max and Team plans, auto mode is on by default and catches risky commands. You can also start tasks from Slack and run it in GitHub Actions.
There is a cost angle too. Running a Claude model as an agent inside Copilot spends AI Credits; doing the same work in Claude Code spends your Claude subscription allowance. If long tasks make up much of your week, the Claude subscription usually goes further.
Can you use Copilot or Claude Code in China?
GitHub's trade controls policy lists the destinations to which Copilot may not be sold or exported, including Cuba, Iran, North Korea, Russia, Belarus and Crimea. Mainland China is not on that list, so under GitHub's policy it is not a restricted destination for Copilot. GitHub's plans page does not say whether every third-party model appears in every region, so check the model list in your own account.
Claude Code is a different story. Anthropic's supported countries list excludes mainland China, Hong Kong and Macau, and the policy covers Claude.ai, Claude Code and the API, so accounts used there risk suspension.
Neither company's pages mention Alipay or WeChat Pay. Developers in China report that GitHub itself connects only intermittently, that Copilot completion and chat usually need a proxy to stay stable, and that payment needs a foreign-currency card such as Visa or Mastercard. Claude's web checkout often declines mainland-issued cards, subscriptions through overseas App Store or Google Play accounts are the more common route, and accounts risk deactivation. This is community experience, not official policy.
For developers in China, that makes Copilot the lower-friction choice of the two.
Using both: how to split the work
The cheapest way to run both is Copilot Pro at $10 plus Claude Pro at $20, for $30 a month. A common split:
Day-to-day coding stays in Copilot. Tab completion, small edits and questions in the editor happen there and do not touch your Claude allowance.
Long tasks go to Claude Code. Cross-directory refactors, bulk dependency upgrades and anything that needs repeated test runs get a short brief with allowed directories and required commands, then Claude Code runs them.
Review stays on GitHub. Changes from both tools arrive as pull requests and get reviewed in one place.
After a month, if long tasks turned out to be rare and most of your Claude Code allowance sat unused, keep only Copilot. If Copilot's AI Credits kept running out and you kept topping up, compare what you spent on top-ups with Claude Pro's $20.
They suit different work. Copilot is better for writing code in your editor, with fast completion, a lower price and a free plan. Claude Code is better for long tasks you hand off. Many developers use both.
How does Claude Code vs Copilot pricing compare?
Copilot has a free plan and Pro costs $10 a month. Claude Code has no free plan and starts with Claude Pro at $20 a month. Heavy-use plans start at $100 a month on both sides.
Can I use Claude models in Copilot?
Yes. The GitHub Copilot plans page lists Claude models on every plan, including Free. Using them spends AI Credits, and the free plan caps chat requests.
Does Copilot work in China?
GitHub's trade controls policy does not list mainland China among Copilot's restricted destinations. GitHub does not say whether every third-party model is available there, so check the model list in your own account.
