The Claude Code vs Codex contest has a clear front-runner: three in four of the 138 developers surveyed by ZDNET’s David Gewirtz use Claude Code in their daily workflows, with Codex maintaining a loyal minority on grounds of cost and workflow simplicity.
The survey drew responses via HARO and Qwoted, two services that connect industry practitioners with journalists. Respondents answered the question: ‘Do you use Claude Code or OpenAI Codex? If so, which did you choose, and why?’ A little more than a third use Codex, and 22% use both tools, which is why the figures exceed 100%.
Why Claude Code Wins on Context and Reasoning
The most-cited reason for choosing Claude Code, named by 14% of respondents, is its ability to hold a large codebase in context so that changes spanning multiple files stay coherent. A further 9% said it plans and reasons through changes before editing, reducing corrections during complex refactors, and another 9% rated its raw code quality and reasoning as better than Codex on equivalent tasks.
Naman Ahuja, a software engineer at Meta, described his approach: ‘I use Claude Code extensively at Meta because it fits naturally into my development workflow. I treat it like a peer: brainstorming ideas, shaping plans, and executing. It has improved my productivity and lets me focus on more business problems to solve. I’m happy with it, though I review its output carefully.’
Mona Rajhans, a senior software engineering manager at Palo Alto Networks, said her entire team uses it daily: ‘What I notice at both levels: it handles context across large, complex codebases better than anything else we’ve tried. Code quality has held up. Adoption was faster than any tool rollout I’ve managed in years.’
The Anthropic Claude Code product page lists features including dynamic workflows across tens to hundreds of parallel subagents, an Agent view for managing sessions (launched 11 May 2026), Routines that run on a schedule or in response to an event (launched 14 April 2026), and computer use allowing Claude to open apps, navigate browsers, and run developer tools (launched 23 March 2026). Anthropic followed those updates by launching Claude Fable 5.1 and Claude Mythos 5.1 on 1 September 2026, describing them as ‘the world’s most advanced models for coding and knowledge work,’ according to the Anthropic Help Center release notes.
Where Codex Holds Its Ground
Cost is Codex’s strongest card. Seven per cent of respondents cited better value for money and the absence of weekly usage caps as their primary reason for choosing it. OpenAI moved Codex from per-message pricing to token-based credit pricing in April 2026: for GPT-5.3-Codex, the rate is $1.75 per million input tokens and $14.00 per million output tokens, according to CloudZero’s analysis of OpenAI Codex pricing. Codex usage limits apply across both a five-hour window and a weekly window, per the OpenAI Help Center, which also confirmed that GPT-5.4 will be replaced by GPT-5.6 Terra in Codex on 31 August 2026.
Chris Seymour, founder of GS Consulting, said: ‘I prefer OpenAI Codex and use it for secure AI and cybersecurity software, testing, research automation, and deployments. I also greatly prefer it over Claude as OpenAI provides much more usage for the same price. I never run out of tokens, and I’m using it every day.’
Adam Dalloul, a computer science student and founder of EmpirioLabs AI (part of the Nvidia Inception programme), who hosts more than 100 AI models on a single API, argued the harness matters as much as the model: ‘My assessment is that as a harness, Codex is far better. The app is much more stable, clearer, and has better QoL features, context management, and compacting.’
Claude Code vs Codex in a Dual-Tool Workflow
One in five respondents uses both tools, often assigning each a distinct role. Michael Rangel, founder of small-business banking platform Novo, said: ‘I use both. Claude Code is my hands-on-desk tool. Codex powers my Hermes agent on the go. Pitting them against each other, with one building and the other reviewing, yields better results than either model alone.’ Rangel co-founded Novo in December 2015 and grew it to serve more than 250,000 small businesses across all 50 US states, raising over $170 million in venture capital, including a $90 million Series B led by Stripes that valued the company at $700 million. Lifetime small business transactions through the platform have passed $12 billion, according to Finovate.
Kai Blakeborough, an enterprise AI strategy and governance adviser, said: ‘Both. At $20/month each, the value is absurd. Claude Code is my visioning and planning tool. Codex handles execution and review. Instead of being locked into one model or harness, I get the best of each, and both improve each other’s outputs.’
Rory Bokser, a Montreal-based developer, takes it further. He uses both at their maximum plans, exhausts weekly usage consistently, and described a regular flow of planning with Claude Fable before handing implementation to Codex, then using both as adversarial reviewers on pull requests.
What the Overall Sentiment Reveals
Beyond tool preference, the survey surfaced broader patterns. Twenty-two respondents run the two tools against each other, with one writing and the other reviewing before anything ships. Fourteen said human review is non-negotiable, treating AI-authored code like a junior developer’s pull request. Six warned about silent failures: code that looks correct and runs while being quietly wrong.
Eighteen respondents described themselves as non-engineers, a group including founders, marketers, a urologist, and a plumber, shipping production software with these tools. Six chose their tool based on trust in the company behind it rather than on benchmarks.
The self-selected nature of the responses means the 138 voices are not a statistically representative cross-section of global developers. What the data does show is a settled hierarchy for now: Claude Code leads on context, reasoning, and trust, while Codex competes hard on cost and workflow integration. The question that will shape the next round of switching is whether Codex’s updated model line, arriving on 31 August 2026, narrows the capability gap enough to change those calculations.