75% of developers I surveyed prefer Claude Code - here's why they choose it over Codex
Anthropic
OpenAI
NVIDIA
Meta
Palo Alto Networks
Salesforce
A survey of 138 developers shows that 75% prefer Claude Code over OpenAI Codex, citing better context handling, reasoning, and code quality. However, 22% use both tools, and some prefer Codex for cost, workflow integration, and autonomy. Human review remains essential regardless of the tool chosen.
In an informal survey of 138 developers, 75% reported using Claude Code, while 22% use both Claude Code and Codex. Reasons for preferring Claude Code include its ability to hold large codebases in context, plan and reason before editing, produce better raw code quality, and work agentically in the terminal. Some developers also cited trust in Anthropic over OpenAI. Conversely, Codex users prefer it for providing more usage for the same price, fitting into existing OpenAI workflows, being predictable and low-maintenance, and running autonomously. Developers who use both tools often run them against each other for review or use one for planning and the other for execution. Overall, respondents emphasized that human review is non-negotiable, and cost and usage limits are common complaints.
- Abbreviations
- ML = Machine Learning — Машинное обучение
- API = Application Programming Interface — Программный интерфейс приложения
- CLI = Command Line Interface — Интерфейс командной строки
- UI = User Interface — Пользовательский интерфейс
- MCP = Model Context Protocol — Протокол контекста модели
- EDA = Exploratory Data Analysis — Разведочный анализ данных
Source: ZDNet AI —
original