What Is Codex? AI Coding Agent By OpenAI That Could Edge Out Software EngineersOpenAI has released a research preview of Codex, an AI coding agent designed to assist software engineers in writing code and addressing bugs.
What Is Codex? AI coder replaces software engineers? AI Coder By OpenAI AI Coding Agent Launched By OpenAI That Could Replace Software Engineers
Codex’s powered by a2 AI reasoning model specifically trained on software engineering tasks.
OpenAI unveiled Codex, the company’s most powerful AI coding tool to date, on Friday (May 16). Accessible to ChatGPT Pro, Enterprise and Team members, the software engineering agent runs in the cloud and can be a “virtual coworker” to engineers, assisting them in writing code, fixing bugs — and the like — at breakneck speed.
The company’s chief executive Sam Altman, had the following to say on social media about the product powered by the o3 reasoning model: “Today we’re opening a research preview of new AI tools, so that everyone can see and understand how AI will create new possibilities for them.
“Today we are announcing Codex. A software engineering agent that lives in the cloud and takes care of things for you, like implementing a new feature or fixing a bug. You can parallel many tasks,” wrote Mr Altman on X (formerly Twitter).
According to OpenAI, Codec is able to “read and edit files, and execute commands, including test harnesses, linters and type checkers”. Codex needs between 1 and 30 minutes to finish the code, based on the difficulty of the tasks.
Codex has been designed to enable users to run concurrent sessions at once, so they can have multiple agents working in parallel.
Also Read : Samsung Galaxy A55 5G gets a flat Rs 12,000 discount on Amazon | Grab Deal
today we are introducing codex.
— Sam Altman (@sama) May 16, 2025
it is a software engineering agent that runs in the cloud and does tasks for you, like writing a new feature of fixing a bug.
you can run many tasks in parallel.
How to use Codex?
In order to start using Codex, users simply need to access the sidebar on ChatGPT.
Give the AI agent another coding task by typing a prompt and then pressing ‘Code’.
While performing a task, the agent is not allowed to access the internet — only the code that is given to the agent explicitly from GitHub repositories.
Upon finishing a task, Codex gives users irrefutable proof of its work in the form of terminal log citations.
When in doubt, or if the tests fail, the Codex agent will indicate it, so the user has the possibility to take the appropriate decision.
Future of software engineers :
Software engineer AI tools have gained popularity over the past few months. Many IT companies have also been saying that with AI at the forefront, writing code can even become an outdated profession. The CEOs of technology giants such as Google and Microsoft have already said that around 30 percent of their companies’ code is now written by AI.
Also Read : Lava Shark 5G India Launch on May 23; Teased to Be Priced Under Rs 10,000
Codex’s release could potentially even hasten the timeline for AI-produced coding. Asked what software engineering would be like in 10 years, the Codex team suggested that speed and reliability of coding could increase, pointing towards more of a reliance on AI.
“We should be able to write down a decent description of some software we want, and get that software, and it also doesn’t take forever, written down, is what the term I would use, to train a new one (and have it be reliable),” Jerry Tworek, a VP of Research at OpenAI wrote in an AMA on RedditSome Dude responded: “Let me translate into plain English: Software engineers better be scared and running to upgrade their skills, like yesterday.
