OpenAI’s 'Rising Star' Thibault Boldly Leads 28-Day Commitment as Head of Codex

If you’ve been keeping an eye on OpenAI’s coding tool Codex lately, you’ve probably come across the name "Tibo" at some point. It is a nickname for Thibault Sottiaux, who serves as the Head of Codex. Recently, as OpenAI has been rolling out a major campaign to significantly boost Codex's performance and usability, the industry spotlight has been turning toward him.

The Current Status of Codex Under Thibault Sottiaux's Leadership

Codex is a coding agent developed in-house by OpenAI and has established itself as a key service that now supports cutting-edge models such as GPT-6.1 Sol and GPT-6 Astra ¹²³. Thibault Sottiaux is the overall lead of this project ⁴. For industry insiders and developers, it is essential to pay attention to his role and activities to understand the direction in which Codex is evolving.

In fact, the combined user base of Codex and ChatGPT Work has rapidly grown, surpassing a staggering 40 million users ¹. Given that he is overseeing such a complex service with such rapid user growth, it is easy to understand why he is emerging as a "rising star" in the industry.

A 28-Day Relay of Promises: Daily Improvements or Usage Resets

The pivotal event that brought Tibo into the public spotlight occurred on October 4, 2026. He promised that for the next 28 days, he would either deliver concrete improvements to Codex users every day or, if that were not possible, reset their entire usage allowances ¹⁵.

Since he made the promise, updates have indeed been rolling out every day. For example, on day 1, performance upgrades were applied to GPT-6 Astra and GPT-6.1 Sol, and on day 4, Codex received an Ultrafast mode ¹. On day 7, the default context length of GPT-6.1 Sol was expanded significantly to 500k tokens ¹.

Why the Industry Is Watching and Specific Achievements

Tibo’s relay campaign was designed to provide tangible benefits to users. Between October 5 and 17, the base speeds of GPT-6 Astra and GPT-6.1 Sol improved by approximately 50%, and the context window increased from 272k to 500k ⁵. Industry analysts interpret this intensive improvement campaign as a strategy to address the usage limit issues experienced by users in a competitive landscape and enhance the service experience ⁵.

The improvement in the model specifications themselves is widely recognized as a clear strength. These technological advancements serve as the foundation for simultaneously elevating the stability and performance of the Codex ecosystem.

The Future of the Codex Ecosystem and Strategic Decisions by the Head

Underneath the Codex led by Thibault Sottiaux lies OpenAI’s extensive AI infrastructure. The GPT-6 Sol on which Codex relies touts "low cost, high efficiency" as its key weapon, having halved the error rate compared to its predecessor. According to ZDNET, GPT-6 Sol delivered performance on par with competitor models (Claude Opus 5) on the AutomationBench benchmark while reducing costs to just 9% ³. Additionally, GPT-6 Sol significantly lowered the error rate for unintentionally exposing private information from 52% in the previous model (GPT-5.6 Sol) to 11%, improving both security and alignment as an intelligent agent ³. For reference, the model used in the previously mentioned mathematics research (²) was GPT-5.6 Sol.

OpenAI is committing significant resources to the fields of mathematics and software engineering. For instance, in research related to the Navier-Stokes equations, Codex utilizing GPT-5.6 Sol served as a core tool, helping to explore directions for complex mathematical proofs ². Thibault’s rapid deployment speed and strategic judgment, as the person directly leading the coding agent, play a crucial role in determining how these high-performance models are delivered to actual users through Codex. As he engages with users by publishing daily improvements for 28 days, the industry continues to watch closely to see how far he will expand the Codex ecosystem.