OpenAI just launched 'Codex', a new AI agent for software engineering
The new Codex agent from OpenAI will be integrated within ChatGPT and aimed at developers
OpenAI has unveiled the launch of a new AI agent designed specifically for software engineering tasks.
Codex, which is now available in preview, allows users to “delegate tasks to a software engineering agent in the cloud”, the company said. The agentic AI tool is powered by codex-1, a version of OpenAI’s o3 reasoning model which is optimized for coding tasks.
During training of codex-1, OpenAI said it aimed to “align outputs closely with human coding preferences and standards”. All told, codex-1 consistently produces "cleaner" code outputs compared to its o3 reasoning model.
“Codex can perform tasks for you such as writing features, answering questions about your codebase, fixing bugs, and proposing pull requests for review; each task runs in its own cloud sandbox environment, preloaded with your repository,” the company said in a blog post.
How OpenAI’s new Codex tool works
The new AI agent is conveniently integrated within ChatGPT, the company explained in a blog post. Users can access Codex through the sidebar in the chatbot and assign it tasks by typing a simple prompt and clicking ‘Code’.
“If you want to ask Codex a question about your codebase, click “Ask”. Each task is processed independently in a separate, isolated environment preloaded with your codebase.”
Codex is capable of reading and editing files, and can run commands such as test harnesses, linters, and type checkers, OpenAI noted. The firm added that task completion typically takes “between 1 and 30 minutes” depending on the complexity of said task.
Sign up today and you will receive a free copy of our Future Focus 2026 report - the leading resource for IT decision-maker insight on priorities and investment areas in AI, security and more.
The tool’s activities can also be monitored in real-time, allowing developers to keep tabs on progress.
Notably, once completed Codex will provide users with "verifiable evidence” of its actions via citations of terminal logs and test outputs.
This, the company explained, allows you to “trace each step taken during task completion”.
“You can then review the results, request further revisions, open a GitHub pull request, or directly integrate the changes into your local environment. In the product, you can configure the Codex environment to match your real development environment as closely as possible.
How to get your hands on the new Codex agent
While Codex is being released in preview initially, it will be rolled out to ChatGPT Pro, Enterprise, and Team users immediately.
In its blog post unveiling the tool, OpenAI said support for Plus and Edu editions will be “coming soon”.
The launch of the new AI agent comes in the wake of OpenAI’s rumored acquisition of Windsurf, a US-based startup specializing in AI coding tools, and highlights the company’s continued push into the AI programming domain.
A host of big tech companies have ramped up expansion of AI tools for software development in the last 18 months, which has become a leading use-case for the technology since the emergence of generative AI in late 2022.
AWS, Microsoft, GitHub, and Google have all expanded AI coding capabilities across their respective product bases.
OpenAI has been equally active on this front, with recent model releases specifically targeting coding capabilities.
The company has been keen to push its GPT-4.1 model as a developers go-to option, but recent benchmark testing raised questions over its effectiveness compared to other leading industry options.
MORE FROM ITPRO
- ‘DIY’ agent platforms are big tech’s latest gambit to drive AI adoption
- OpenAI’s day in the sun is over
- OpenAI unveils its Operator agent to help users automate tasks – here's what you need to know
Ross Kelly is ITPro's News & Analysis Editor, responsible for leading the brand's news output and in-depth reporting on the latest stories from across the business technology landscape. Ross was previously a Staff Writer, during which time he developed a keen interest in cyber security, business leadership, and emerging technologies.
He graduated from Edinburgh Napier University in 2016 with a BA (Hons) in Journalism, and joined ITPro in 2022 after four years working in technology conference research.
For news pitches, you can contact Ross at ross.kelly@futurenet.com, or on Twitter and LinkedIn.
-
OpenAI pens AI safety agreement with four firmsNews Under pressure after rogue AI incidents, OpenAI details plans for external evaluators
-
OpenAI says its new cyber-focused AI model found thousands of Linux kernel vulnerabilitiesNews The new cyber-focused AI model boasts improved bug hunting capabilities compared to its predecessor
-
OpenAI won't release next Astra model over safety worriesNews AI developer says new agent didn't stay within the scope of tasks
-
OpenAI says some researchers are blowing through $7,000 in AI tokens every day – but it’s a price the company appears willing to payNews OpenAI has revealed that researchers now spend around $600 each day on AI tokens amidst a surge in agentic coding. Some, meanwhile, are using upwards of $7,000 worth of tokens per day.
-
Six things OpenAI learned about AI from the Hugging Face incidentNews OpenAI's report into AI going rogue reveals efforts at cheating and communicating — but also some well-behaved bots
-
OpenAI forges closer ties with IBM in enterprise pushNews The duo will combine OpenAI models and products with IBM Consulting expertise
-
OpenAI has paused work on its Astra AI model after it passed a 'critical threshold' in cyber capability – but it’s not the one that breached Hugging FaceNews The firm said it's Astra model can "develop functional zero-day exploits of all severity levels"
-
AI testing firm Irregular the source of ‘misconfigurations’ that led to Meta, OpenAI, and Anthropic AI incidentsNews The “frontier security lab” has been referenced in multiple cyber incident statements