OpenAI’s annual developer conference on Monday was a spectacle of ambitious AI product launches App Store for ChatGPT for a surprise Video-Generation API Who brought creative concepts to life. But for enterprises and tech leaders watching closely, the most consequential announcement was the quiet one. General availability of codecsThe company’s AI software engineer. This release signals a profound change in the way software—and by extension, modern business—is built.
While other announcements caught the public’s attention, the production-ready release manualsupercharged by a new special model and a Suite of enterprise-grade toolsis the engine behind OpenAI’s entire approach. It is the tool that makes the tool, the proven agent in a world full of agentive capabilities, and the clearest expression of a company’s strategy for winning the enterprise.
General availability of codecs takes it from a "Research Preview" For a fully supported product, complete with a new Software Development Kit (SDK)A slack integrationand administrative controls for security and monitoring. This change announces that Codex is ready for mission-critical work inside the world’s largest companies.
"We believe this is the best time in history to be a builder; Going from idea to product has never been faster," OpenAI CEO Sam Altman said during this inaugural keynote speaker presentation. "Software took months or years to develop. You see, building with AI can now take minutes."
That acceleration is not theoretical. This is a reality born out of OpenAI’s own internal use – a huge "dog food" Effort that serves as the ultimate case study for enterprise customers.
Inside GPT-5-Codex: The AI model that codes autonomously for hours and delivers 70% productivity gains
The codec is at the heart of the upgrade GPT-5-CodexWhich is a version of OpenAI’s latest flagship model "Deliberately trained for codecs and agentive coding." The new model is designed to act as an autonomous teammate, going far beyond simple code autocompletion.
"I personally like to think of it like a human teammate," OpenAI engineer Tibo Sotiaux explained during a technical session on Codex. "You can associate it with any program on your computer, you can delegate it, or as you will see, you can give it tasks without explicit prompting."
This new model enables "adaptive thinking," This allows the time and computational effort spent on a task to be dynamically adjusted depending on its complexity. For simple requests, it is fast and efficient, but for complex refactoring projects, it can take hours.
An engineer during the technical session said, "I have seen the GPT-5-codecs model work productively for more than seven hours in a marathon session." This ability to handle long-running, complex tasks is a significant leap beyond the simple, single-shot interactions that define most AI coding assistants.
The results inside OpenAI have been dramatic. The company reported that 92% of its technical staff now use codecs daily, and those engineers complete 70% more pull requests (a measure of code contribution) each week. Usage has increased tenfold since August.
"It feels great when we look at the stats as a team," Sotiax shared. "But even better is having lunch with someone who then says, ‘Hey, I use Codex all the time. Here’s a cool thing I do with it. Do you want to hear about it?’"
How OpenAI uses Codex to build its own AI products and catch hundreds of bugs every day
Perhaps the most compelling argument for the importance of Codex is that it is the foundational layer upon which OpenAI’s other exciting announcements were built. during god day eventThe company showcased custom-built arcade games and a dynamic, AI-powered website for the conference, which was developed using all manual,
In one session, engineers demonstrated how they built "storyboard," A custom creative tool for the film industry in just 48 hours during an internal hackathon. "We decided to test our coding agent Codex… We would send tasks to Codex between meetings. We really easily reviewed PRs and merged into production, which Codex allowed us to do even from our phones," said Alison August, solutions engineering leader at OpenAI.
This reveals an important insight: the rapid innovation demonstrated at Devde is a direct result of the productivity flywheel created by manualAI is a core part of the creation process of all other AI products.
A key enterprise-focused feature is the new, more robust code review capability. OpenAI said this "GPT-5-codecs intentionally trained to excel at highly thorough code reviews," This enables it to validate the programmer’s intent against the actual implementation to trace dependencies and find high-quality bugs. Internally, almost every pull request at OpenAI is now reviewed by Codex, which catches hundreds of issues per day before they reach a human reviewer.
"This saves you time, lets you ship with more confidence," Sotiaux said. "There’s really nothing worse than finding a bug after you’ve actually shipped a feature."
Why are enterprise software teams choosing Codex over GitHub Copilot for mission-critical development?
maturity of manual It is central to OpenAI’s broader strategy to conquer the enterprise market, a necessary step to justify its huge valuation and unprecedented compute expenditure. CEO Sam Altman confirmed the strategic change during a press conference.
"The models exist now, and you should expect a massive focus from us on truly winning enterprises with amazing products, starting from here." Altman said during a private press conference.
Greg Brockman, president and co-founder of OpenAI, was quick to say, "And you can see it already with Codex, which I think has been an incredible success and has grown really quickly."
For technology decision makers, the message is clear. While consumer-facing agents who book dinner reservations are still holding their own, manual is a proven enterprise agent that is providing substantial ROI today. Companies like Cisco have already introduced Codex into their engineering organizations, cutting code review time by 50% and reducing project timelines from weeks to days.
with new codecs sdkCompanies can now embed this agentic power directly into their own custom workflows, such as automating fixes in a CI/CD pipeline or even building self-developed applications. During a live demo, an engineer showed off a mobile app that updated its own user interface in real-time based on natural language prompts, all powered by the embedded codecs SDK.
While the launch of one App ecosystem in ChatGPT and breathtaking views of sora 2 api correctly crafted headlines, General availability of codecs This marks a more fundamental and immediate change. It is the quiet but powerful engine driving the next era of software development, turning the abstract promise of AI-powered productivity into a concrete, deployable reality for businesses today.
