Companies test Codex, but Claude Code remains the preferred choice among engineers

Across enterprise engineering teams evaluating the latest generation of AI coding agents, Anthropic’s Claude Code has emerged as the tool developers actually want to use. OpenAI’s Codex is getting its fair share of trials, but when engineers get a vote, Claude Code keeps winning the room.

The numbers tell the story

By February 2026, Claude Code was responsible for roughly 4% of all public GitHub commits, averaging around 135,000 commits per day. That number spiked to a daily peak of 326,000 commits by mid-March 2026.

On verified benchmarks, Claude Code posted a 72.7% score compared to Codex’s 69.1%.

Developer reports throughout 2026 have consistently highlighted Claude Code’s ability to manage long-session contexts, the kind of extended, multi-step coding tasks where an AI agent needs to remember what it did 40 steps ago without losing the thread. It also gets high marks for UI-related tasks and its plugin ecosystem.

Codex holds an advantage in one area: price-per-task efficiency. For certain delegation-style tasks where you’re handing off well-scoped, relatively contained work, Codex can be the more economical option.

Enterprise adoption is shifting

Anthropic reportedly surpassed OpenAI on enterprise adoption metrics starting in early 2025, specifically in developer satisfaction and agentic task completion rates.

Claude Code is gaining traction in enterprise contracts, particularly in environments where consistent output quality and compliance matter. Financial services, healthcare tech, and regulated industries are gravitating toward Anthropic’s offering.

Several firms are quietly adopting hybrid workflows that deploy both tools depending on the task at hand. Claude Code handles the complex, context-heavy work. Codex picks up the more routine, cost-sensitive tasks.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.