Survey: 75% of developers prefer Claude Code over Codex

ZDNET's survey of 138 developers reveals a clear preference for Anthropic's Claude Code over OpenAI's Codex in daily AI coding workflows. Three out of four respondents use Claude Code, while just over a third use Codex, with 22% using both tools. The responses, gathered via HARO and Qwoted, echo broader industry sentiment: developers increasingly favor Claude Code for its agentic, multi-file reasoning capabilities and its ability to maintain coherence across large, complex codebases.

Key reasons emerged for this preference. Fourteen percent of respondents highlighted Claude Code's superior context handling, which keeps changes consistent across many files. Nine percent praised its planning and reasoning before edits, reducing corrections during complex refactors, while another 9% reported better raw code quality than Codex on identical tasks. Developers from major tech firms backed these claims: a Meta engineer described using it as a peer for brainstorming and execution, an Nvidia data scientist valued that it reads the entire repo before editing, and a Palo Alto Networks manager noted adoption was faster than any tool rollout in years. Despite Claude Code's dominance, the survey underscores that human review and final judgment remain critical—AI accelerates routine work but can't replace architectural oversight.