Who Is Thibault Sottiaux, OpenAI Head of Product and Codex Lead?

Thibault Sottiaux, who goes by Tibo, leads OpenAI's core products, including Codex, as of October 2026.

He told TechCrunch that he leads "all of core products," which he listed as the API, agent infrastructure, enterprise, all of ChatGPT (including ChatGPT Work and ChatGPT classic), and Codex.

Before that, Sottiaux led the Codex engineering team. He worked at DeepMind in London before OpenAI. There, he led the Gemini human data team for a time.

Who Thibault Sottiaux Is

Thibault Sottiaux leads the Core Products and Platform organization at OpenAI, managing Codex alongside ChatGPT and developer infrastructure.

Sottiaux's official title evolved alongside OpenAI's organizational structure throughout 2026. In January 2026, Cerebral Valley and Software Engineering Daily identified him as the engineering lead of Codex. He managed a dedicated unit of roughly 30 engineers. By February 2026, host Dan Shipper on the Every podcast introduced him as the head of Codex.

By summer 2026, OpenAI widened his operational remit. In an August 2026 interview, TechCrunch called him OpenAI's head of product. In September 2026, The Pragmatic Engineer similarly reported that Sottiaux heads the Core Products and Platform organization at OpenAI.

• Current role: Leads OpenAI's core products, including Codex (previously Codex engineering lead)
• Organization: OpenAI, reporting to cofounder and president Greg Brockman
• Known for: Helping create Codex and resetting Codex users' token limits at growth milestones
• In the news: An October 4, 2026 pledge to ship either a clear improvement or a full reset each day for 28 days

Early Career and Research at Google DeepMind

Before OpenAI, Thibault Sottiaux ran a supply-chain startup, worked at Google in London, and then joined DeepMind.

In a January 2026 interview with Cerebral Valley, Sottiaux recounted that he dreamed about speaking directly to computers during his childhood. He studied artificial intelligence at university. The interview does not name the university or his degree. He then ran a supply-chain startup before joining Google in London.

At Google, Sottiaux spent two years working in the advertising division. A visiting executive from California abruptly cancelled the project, as reported by The Pragmatic Engineer. Sottiaux subsequently moved to Google DeepMind in London. DeepMind employed roughly 400 people during those formative years. At DeepMind, Sottiaux was among a group of researchers who hypothesized that scaling language models could lead toward artificial general intelligence (AGI). The group's internal "chat with an LLM" project spread quickly across DeepMind. DeepMind never launched it as an external commercial product.

Sottiaux later led the human data team for Gemini. That operation scaled to several hundred contributors ahead of the initial Gemini model release. He is listed as a co-author on three DeepMind-era papers: Reverb: A Framework For Experience Replay (February 2021), the Gopher language model paper (December 2021), and the Gemini technical report (December 2023).

Transition to OpenAI and Early Codex Work

Thibault Sottiaux joined OpenAI in 2024. The Pragmatic Engineer calls him one of the engineers who created Codex.

According to The Pragmatic Engineer, Sottiaux decided to seek an opportunity at OpenAI after discovering in 2023 that a team of roughly 20 engineers had built and maintained ChatGPT. He told Cerebral Valley, "I want to join something that feels closer to my values, and where the talent density is higher." He joined "just before reasoning models, before the o1 model."

Sottiaux is a listed author on the OpenAI o1 System Card. The card was posted on arXiv in December 2024. At OpenAI he first built "tools primarily to help researchers move faster." He then built prototypes "that were pre-Codex." Codex first shipped as Codex Web. He described Codex Web as "fully in the cloud and uninterruptible, going from prompt to pull request in one go."

Codex Technical Architecture and Internal Adoption

Codex grew from a command-line interface (CLI) tool with around 50 early users into a coding system used by about 95 percent of OpenAI's software engineers.

In his interview with Cerebral Valley, Sottiaux noted that early development of Codex was split across two parallel tracks, including a terminal client that had approximately 50 users immediately after its initial release. By January 2026, Sottiaux reported that roughly 95 percent of software engineers at OpenAI relied on Codex. It reviewed 100 percent of internal code changes. One person built the Codex integrated development environment (IDE) extension. Sottiaux said "there were more than 30 PRs per day being generated and merged" and that "The IDE extension has been our most successful surface."

When OpenAI launched the Codex app in February 2026, the software surpassed one million downloads within its first week, as Sottiaux discussed on the Every podcast. Sottiaux said the Codex Super Bowl ad brought "a giant surge of traffic" very quickly after it aired at 4:00 p.m. PST. The engineering team rebuilt the server stack on WebSockets using persistent connections. Sottiaux said that cut overall turn latency by "something like 30 to 40 percent." The fast model he discussed "is powered by Cerebras."

The engineering team chose to write Codex in Rust because they anticipated instances running across millions of cloud machines. Performance, security, and efficiency came first. Codex was released as open-source software capable of working with third-party language models rather than being restricted to OpenAI's proprietary models. Sottiaux acknowledged to The Pragmatic Engineer that competitors copying Codex features could sting. He noted that it is the price of working in the open.

Product Strategy and Executive Responsibilities

Sottiaux's remit grew in a 2026 OpenAI reorganization.

Reporting from WIRED in May 2026 revealed that OpenAI cofounder and president Greg Brockman assumed leadership over overall product strategy to unify ChatGPT and Codex into a coordinated user experience. In July 2026, Daring Fireball quoted reporting that Sottiaux "has been tapped to lead the company's core product and platform teams." The same text called him "a key leader" in building Codex into "one of the company's fastest-growing products of all time." That reporting named Sottiaux as one of the leaders overseeing development of OpenAI's planned desktop super app. That initiative is intended to consolidate Codex, ChatGPT, and the Atlas web browser into a single desktop interface.

In August 2026, Sottiaux spoke to TechCrunch. He confirmed that he reports directly to Brockman. He remarked that everyone at OpenAI ultimately reports to Greg. He also said ChatGPT Work had hit 20 million users and is included in the Plus plan, "which is only $20 a month." TechCrunch wrote that users "might know Thibault Sottiaux as the guy who resets their token limits whenever the product hits a growth milestone."

Recent Announcements and the October 2026 Pledge

In October 2026, Thibault Sottiaux made a public commitment to ship daily improvements or usage resets for Codex and ChatGPT Work users.

On October 4, 2026, Sottiaux posted on X: "Over the next 28 days, each day we'll either ship one thing that is a clear improvement and relevant for most codex/work users or ship a full reset. Let the improvements begin." Nerdschalk reported on the post. Approximately 16 hours prior to announcing the 28-day sprint, Sottiaux said the team was "locking in" on simpler products, more efficient usage, features, and models. He gave no end date. The 28 days run to October 31 or November 1. As of October 10, 2026, no readable report confirms which improvements or resets have shipped.

The pledge followed major platform announcements delivered at OpenAI DevDay 2026 in early October. According to coverage from InfoQ, OpenAI revealed that Codex can execute tasks in cloud environments alongside local developer machines. The Codex CLI got voice input and an /agents interface. InfoQ also reported a code-review workflow for GitHub pull requests alongside GitLab merge requests. Codex Security Cloud scans repositories and new commits. GPT-6.1 Sol is available through the API, ChatGPT Work, and Codex.

Engineering Philosophy and Developer Perspectives

Sottiaux has said the clearest bottleneck in AI-assisted coding is how fast people can verify that code is correct.

On the Every podcast with Dan Shipper, he said, "The bottleneck that's very apparent is how fast you can verify that things are correct," and added, "You cannot delegate understanding." He told Cerebral Valley in January 2026, "If you only focus on code generation, you aren't going to be moving fast enough, I'd say, in 2026." He also spoke on Software Engineering Daily. He said, "I feel it's an amazing time to have problems as solving them has never been easier."

In the same January 2026 Cerebral Valley interview, Sottiaux predicted that "agents will become so reliable that they will be able to run for days, if not weeks." He explained that his team optimizes for return on investment. They measure that investment specifically in human attention. On hiring, he said, "I look for the most AGI-pilled individuals in the world." The Pragmatic Engineer reports his tip for people joining the Codex team: "have you asked Codex?"

He told TechCrunch, "The mission of OpenAI is to bring everyone along." He added, "Our goal is to, over time, include more utility in the same dollar amount." He also said, "It's important to pick models that are safe and aligned."

What Is Not Published About Sottiaux

Thibault Sottiaux's age, net worth, and compensation are not published.

Sottiaux has referenced studying artificial intelligence at university and founding a supply chain startup prior to joining Google, but neither the university nor the startup is named in those interviews.

His LinkedIn profile is linked from The Pragmatic Engineer's episode page. Sottiaux has posted about Codex on X. He did so with the October 4 pledge.