Analysis

Developers are leaving Copilot and editor plugins for Claude Code and Codex

In six months Claude Code went from 18% to 39% of professional developers and Codex from 3% to 16%, while Copilot and Cursor fell. Price, tools and computer use explain it.

AI Fin ResearchCovers Global
The editor was built for typing code. The agent types it now.
39%of professional developers using Claude Code at work in May to July 2026, up from 18% in January
16%using Codex, up from 3% in January
21%using GitHub Copilot, down from 29% a year earlier

Two tools grew and two shrank

JetBrains, which makes editors, surveyed more than 15,000 professional developers for its Developer Ecosystem Survey 2026.

Tool Earlier May to July 2026
Claude Code 18% in January 39%
Codex 3% in January 16%
GitHub Copilot 29% a year earlier 21%
Cursor 18% in January 12%

JetBrains’ own summary: “GitHub Copilot has lost its leadership.” The two tools that grew are made by Anthropic and OpenAI, the companies that train the models inside them.

The lab’s own plan is the cheap one

Tool Plan Price a month How use is charged
Claude Code Pro 20 US dollars, or 17 billed yearly “Claude Code included”
Claude Code Max From 100 “Choose 5x or 20x more usage than Pro”
Codex ChatGPT Plus 20 An allowance of messages every five hours
Codex ChatGPT Pro From 100 “Pro plans currently have no five-hour limit”
GitHub Copilot Pro 10 1,500 AI credits a month
GitHub Copilot Pro+ 39 7,000 AI credits a month
Any tool, with an API key Pay as you go Opus 5.5 at 4 US dollars a million input tokens and 20 a million output By the token

Prices are from each company’s page on 9 October 2026.

A lab that owns the model can sell its agent on a flat plan. A plugin maker buys the model and passes a meter on to you. One full run of a ten-stage paper pipeline uses about 200,000 input and 100,000 output tokens, which costs 2.80 US dollars at Opus 5.5 API prices. A subscriber on a flat plan runs it again and again inside the plan. Token management covers the arithmetic.

New abilities ship in the labs’ own tools

Shipped in Claude Code When
Auto mode Week of 23 March 2026
Computer use in the terminal Week of 30 March 2026
/goal Week of 11 May 2026
Auto mode as the default 14 August 2026

Anthropic’s pricing page adds that “Claude Cowork is now just Claude,” rolling out to Pro and Max plans, so its second agent comes with the same subscription.

Computer use takes the editor out of the loop

An editor plugin sees the file you have open. On the OSWorld benchmark of real desktop tasks, the top agent completed 86.0% on 9 October 2026 against a human baseline of 72.36%. An agent that can run the terminal, the browser and the spreadsheet needs no integration with an editor to do the work.

The editors now host the agents

VS Code’s documentation, dated 7 October 2026, describes agents that “find relevant code, make changes, and run checks without you directing each search, edit, and test run,” and lists Anthropic’s Claude and OpenAI’s Codex among the agents it can run. The Claude Code plugin for JetBrains editors “requires the Claude Code CLI, installed separately.”

The editor has become a window onto an agent made by someone else. Claude Code explained shows what you get by going to the agent directly, and the SEC filings guide is a first project that needs no editor at all.

Sources

Related

Analysis

Throw out what you knew about AI before September

Fable 5.1, GPT-6 Astra and Opus 5.5 arrived inside five weeks. One science benchmark doubled in two months. A test you ran in the spring describes a different technology.

Anthropic, Introducing Claude Opus 5.5Global