Two of the largest developer surveys of the year came out within weeks of each other, and they tell the same story. Claude Code, Anthropic's coding agent, has become the AI coding tool developers use most, overtaking GitHub Copilot, which led the market for years. The surveys also show what developers actually do with these tools, and what makes them switch.

JetBrains: Claude Code doubled in six months

JetBrains asked more than 15,000 professional developers worldwide between May and July 2026 which AI coding tools they use at work. The answers, published in August as part of its Developer Ecosystem Survey 2026:

Claude Code: 39%, up from 18% in January. In the US the figure is 47%. For 31% of developers it is the tool they use most.
GitHub Copilot: 21%, down from 29% a year earlier.
OpenAI Codex: 16%, up from 3% in January, the fastest growth in the survey.
Cursor: 12%, down from 18% in January.
JetBrains AI: about 9%, OpenCode: 7%, Google Antigravity: 6%.

Almost everyone now works with agents: 90% of professional developers used an AI coding agent at work at least once a week, and 68% used one every day.

Stack Overflow: the same leader, and the reasons behind it

Stack Overflow published its 16th annual Developer Survey on October 6. More than 30,000 people took part over seven weeks. Among coding agents, the leaders are Claude Code at 66% and GitHub Copilot at 59%. On AI models, Stack Overflow says Anthropic and OpenAI are neck and neck, with Google a distant third.

Why are the percentages so different from JetBrains'? The surveys ask different questions of different groups. JetBrains counts all professional developers and asks what they use at work; Stack Overflow's figures describe its coding-agent question, where one person can name several tools. The numbers can't be compared directly, but the order is the same in both: Claude Code first, Copilot second.

The most useful part of the Stack Overflow data is why people change tools. Of 13,144 respondents:

33.7% haven't switched AI tools.
25.1% switched for better output quality.
10.6% switched because their employer required or recommended another tool.
7.4% switched for better integration with their workflow.
6.4% switched for a lower price.

The same priority shows up elsewhere: when choosing prompts and models, 75% of developers say quality matters, and 38% say cost does.

What developers use AI for

The most common tasks, among 12,547 respondents: writing code in an area they already know (69%), debugging and refactoring (64%), answering straightforward technical questions (59%), writing tests (58%) and writing code in an unfamiliar area (56%).

Trust has limits. Only 48% trust AI when they can easily check the answer, and 93% say they need to see the sources behind an answer before trusting it.

What this means if you're choosing a coding tool

If you want an agent that takes on whole tasks, such as changing several files, running tests and fixing what breaks, Claude Code is now the most common choice among professionals. OpenAI's Codex is the fastest-growing alternative.

If your company already pays for GitHub, GitHub Copilot is still the second most-used tool, and about one developer in ten changed tools because their employer told them to. Compare the two side by side: Claude Code vs GitHub Copilot.

If you prefer to work inside an AI-first editor, Cursor lost share this year but is still used by one professional developer in eight: Cursor vs Copilot.

If you want open source, OpenCode reached 7%: OpenCode vs Claude Code.

The main lesson from both surveys: developers pick tools by the quality of the result, not by price. Try two or three tools on a real task from your own project before committing, and check the code they write before you ship it.