OpenAI launches preview of Codex AI SWE agent for developers

Shock! Simply days after reviews emerged suggesting OpenAI was shopping for white-hot coding startup Windsurf, the previous firm seems to be launching its personal competitor service as a analysis preview underneath its model title Codex, going head-to-head in opposition to Windsurf, Cursor, and the rising record of AI coding instruments provided by startups and huge tech corporations together with Microsoft and Amazon.

Not like OpenAI’s earlier Codex code completion AI mannequin, the brand new model is a full cloud-based AI software program engineering (SWE) agent constructed atop a fine-tuned model of OpenAI’s o3 reasoning mannequin that may execute a number of improvement duties in parallel.

Beginning immediately it will likely be out there for ChatGPT Professional, Enterprise, and Workforce customers, with help for Plus and Edu customers anticipated quickly.

Codex’s evolution: from mannequin to autonomous AI coding agent

This launch marks a major step ahead in Codex’s improvement. The unique Codex debuted in 2021 as a mannequin for translating pure language into code out there by means of OpenAI’s nascent software programming interface.

It was the engine behind GitHub Copilot, the favored autocomplete-style coding assistant designed to work inside IDEs like Visible Studio Code.

That preliminary iteration centered on code technology and completion, educated on billions of traces of public supply code.

Nevertheless, the early model got here with limitations. It was vulnerable to syntactic errors, insecure code recommendations, and biases embedded in its coaching information. Codex sometimes proposed superficially right code that failed functionally, and in some instances, made problematic associations based mostly on prompts.

Regardless of these flaws, it confirmed sufficient promise to ascertain AI coding instruments as a quickly rising product class. That unique mannequin has since been deprecated and became the title of a brand new suite of merchandise, in response to an OpenAI spokesperson.

GitHub Copilot formally transitioned off OpenAI’s Codex mannequin in March 2023, adopting GPT-4 as a part of its Copilot X improve to allow deeper IDE integration, chat capabilities, and extra context-aware code recommendations.

Agentic visions

The brand new Codex goes far past its predecessor. Now constructed to behave autonomously over longer durations, Codex can write options, repair bugs, reply codebase-specific questions, run checks, and suggest pull requests—every process working in a safe, remoted cloud sandbox.

The design displays OpenAI’s broader ambition to maneuver past fast solutions and into collaborative work.

Josh Tobin, who leads the Brokers Analysis Workforce at OpenAI, stated throughout a current briefing: “We consider brokers as AI methods that may function in your behalf for an extended time period to perform huge chunks of labor by interacting with the actual world.” Codex matches squarely into this definition. “Our imaginative and prescient is that ChatGPT will turn out to be virtually like a digital coworker—not simply answering fast questions, however collaborating on substantial work throughout a spread of duties,” he added.

Figures launched by OpenAI present that the brand new Codex-1 SWE agent outperforms all of OpenAI’s newest reasoning fashions on inside SWE duties.

New capabilities, new interface, new workflows

Codex duties are initiated by means of a sidebar interface in ChatGPT, permitting customers to immediate the agent with duties or questions.

The agent processes every request in an air-gapped setting loaded with the consumer’s repository and configured to reflect the event setup. It logs its actions, cites check outputs, and summarizes adjustments—making its work traceable and reviewable.

Alexander Embiricos, head of OpenAI’s Desktop & Brokers crew (and the previous CEO and co-founder of screenshare collaboration startup Multi that OpenAI acquired for an undisclosed sum final 12 months) stated in a briefing with journalists that “the Codex agent is a cloud-based software program engineering agent that may work on many duties in parallel, with its personal laptop to run safely and independently.”

Internally, he stated, engineers already use it “like a morning to-do record—fireplace off duties to Codex and return to a batch of draft options able to evaluate or merge.”

Codex additionally helps configuration by means of AGENTS.md recordsdata—project-level guides that educate the agent easy methods to navigate a codebase, run particular checks, and comply with home coding types.

“We educated our mannequin to learn code and infer type—like whether or not or to not use an Oxford comma—as a result of code type issues as a lot as correctness,” Embiricos stated.

Safety and sensible use

Codex executes duties with out web entry, drawing solely on user-provided code and dependencies. This design ensures safe operation and minimizes potential misuse.

“That is greater than only a mannequin API,” stated Embiricos. “As a result of it runs in an air-gapped setting with human evaluate, we can provide the mannequin much more freedom safely.”

OpenAI additionally reviews early exterior use instances. Cisco is evaluating Codex for accelerating engineering work throughout its product traces. Temporal makes use of it to run background duties like debugging and check writing. Superhuman leverages Codex to enhance check protection and allow non-engineers to counsel light-weight code adjustments. Kodiak, an autonomous car agency, applies it to enhance code reliability and achieve insights into unfamiliar stack elements.

OpenAI can be rolling out updates to Codex CLI, its light-weight terminal agent for native improvement. The CLI now makes use of a smaller mannequin—codex-mini-latest—optimized for low-latency modifying and Q&A.

The pricing is ready at $1.50 per million enter tokens and $6 per million output tokens, with a 75% caching low cost. Codex is at the moment free to make use of throughout the rollout interval, with price limits and on-demand pricing choices deliberate.

Does this imply OpenAI IS NOT shopping for Windsurf? *Considering face emoji*

The discharge of Codex comes amid elevated competitors within the AI coding instruments area—and indicators that OpenAI is intent on constructing, relatively than shopping for, its subsequent part of merchandise.

Based on current information from SimilarWeb, visitors to developer-focused AI instruments has surged by 75% over the previous 12 weeks, underscoring the rising demand for coding assistants as important infrastructure relatively than experimental add-ons.

Stories from TechCrunch and Bloomberg counsel OpenAI held acquisition talks with fast-growing AI dev software startups Cursor and Windsurf. Cursor allegedly walked away from the desk; Windsurf reportedly agreed in precept to be acquired by OpenAI for a worth of $3 billion, although no deal has been formally confirmed by both OpenAI or Windsurf.

Simply yesterday, in actual fact, Windsurf debuted its family of coding-focused basis fashions, SWE-1, purpose-built to help the complete software program engineering lifecycle, from debugging to long-running venture upkeep. SWE-1 fashions have been reported customized made, educated fully in-house utilizing a brand new sequential information mannequin tailor-made to real-world improvement workflows.

Many issues could also be taking place behind the scenes between the 2 corporations, however to me, the timing of Windsurf launching its personal coding basis mannequin — as a substitute of its technique to-date of utilizing Llama variants and giving customers the choice to fit in OpenAI and Anthropic fashions — adopted in the future later by OpenAI releasing its personal Windsurf competitor, appears to counsel the 2 are usually not aligning quickly.

However however, the truth that this new Codex AI SWE agent is in “analysis preview” to start out could also be a type of OpenAI pressuring Windsurf or Cursor or anybody else to come back to the bargaining desk and strike a deal. Requested in regards to the potential for a Windsurf acquisition and reviews of 1 thereof, an OpenAI spokesperson informed VentureBeat that they had nothing to share on that entrance.

In both case, Embiricos frames Codex as excess of a mere code software or assistant.

“We’re about to bear a seismic shift in how builders work with brokers—not simply pairing with them in actual time, however totally delegating duties,” he stated. “The primary experiments have been simply reasoning fashions with terminal entry. The expertise was magical—they began doing issues for us.”

Constructed for dev groups, not merely solo devs

Codex is designed with skilled builders in thoughts, however Embiricos famous that even product managers have discovered it useful for suggesting or validating adjustments earlier than pulling in human SWEs. This versatility displays OpenAI’s technique of constructing instruments that increase productiveness throughout technical groups.

Trini, an engineering lead on the venture, summarized the broader ambition behind Codex: “This can be a transformative change in how software program engineers interface with AI and computer systems typically. It amplifies every particular person’s potential.”

OpenAI envisions Codex because the centerpiece of a brand new improvement workflow the place engineers assign high-level duties to brokers and collaborate with them asynchronously. The corporate is constructing towards deeper integrations throughout GitHub, ChatGPT Desktop, difficulty trackers, and CI methods. The long-term objective is to mix real-time pairing and long-horizon process delegation right into a seamless improvement expertise.

As Josh Tobin put it, “Coding underpins so many helpful issues throughout the financial system. Accelerating coding is a very high-leverage technique to distribute the advantages of AI to humanity, together with ourselves.”

Whether or not or not OpenAI closes offers for opponents, the message is obvious: Codex is right here, and OpenAI is betting by itself brokers to steer the subsequent chapter in developer productiveness.