Inside Sam Altman's code red response to Gemini 3 - and what it means for ChatGPT

Key Details

Reports say Altman raised the priority level for ChatGPT upgrades from "code orange" to "code red." The memo instructs teams to focus on personalization, speed, reliability, support for more question types, and more consistent day-to-day quality in user conversations.

• According to these reports, some OpenAI staff can temporarily move onto ChatGPT workstreams to support the new push.
• OpenAI has reportedly set up daily calls for leaders responsible for ChatGPT quality improvements.
• A new reasoning model is expected to launch next week, according to the memo, though no public announcement exists yet.
• Advertising tests inside ChatGPT are on hold while the company focuses on experience quality, The Information has reported.
• Projects described as AI agents for shopping and healthcare are also delayed, alongside work on ChatGPT Pulse.
• The memo warns of temporary economic headwinds and urges faster shipping of improvements, according to quoted language.
• OpenAI has not publicly commented on the reported memo, the articles note.

The priority shift affects several previously announced initiatives and keeps ChatGPT product work at the center of OpenAI's near-term schedule. Reporters describe the move as the highest urgency level in the company's color-coded system.

Background Context

Google recently introduced its Gemini 3 family of models, including a version called Gemini 3 Deep Think. Google's official materials say Gemini 3 Deep Think scores higher than earlier Gemini systems on challenging reasoning evaluations. The company highlights results on tests such as Humanity's Last Exam, described by its maintainers as a frontier-level assessment.

Humanity's Last Exam is run by the Center for AI Policy and Scale AI, according to the project site. Scores for public models appear on Scale's official leaderboard and other independent tracking dashboards such as Artificial Analysis. Google's Gemini 3 Pro benchmark documentation lists higher Humanity's Last Exam scores than several competing systems, including OpenAI models.

According to the new reports, OpenAI leadership now sees Google's progress with Gemini 3 as a serious competitive challenge. Altman's memo is framed as a direct response to that progress and to recent evaluation results. The focus remains on improving ChatGPT's everyday performance for users rather than launching new categories of products.

ChatGPT head Nick Turley recently shared usage figures in a LinkedIn post marking three years since ChatGPT's release. He wrote that ChatGPT is the top AI assistant worldwide. According to his post, it accounts for around 70% of assistant usage and about 10% of search activity.

Separate reporting by the Financial Times has estimated OpenAI's weekly active user base at more than 800 million people. Those figures place most users on the free tier, with paid products representing a smaller segment. Reports note that Gemini's user base is growing from a lower starting point.

Source Citations

The following primary sources contain the official statements and technical information referenced above. They include company documentation, project descriptions, leaderboards, independent evaluations, and direct posts from executives.

• Coverage of the reported internal memo in The Wall Street Journal.
• Additional reporting on delayed ads and agent projects in The Information.
• Google's Gemini 3 overview and launch details.
• Gemini 3 Pro benchmark documentation with technical details and evaluation scores.
• The Humanity's Last Exam project description by the Center for AI Policy.
• Scale's official leaderboard for Humanity's Last Exam results.
• Independent benchmark tracking at Artificial Analysis.
• Nick Turley's ChatGPT usage figures on LinkedIn.
• User base estimates for OpenAI reported by the Financial Times.