Sam Altman has reportedly declared a “code red” internal emergency at OpenAI after Google’s Gemini 3 model surged ahead on performance and popularity, prompting the company to pause several projects and refocus on improving ChatGPT. The move reflects concern that OpenAI’s lead in consumer AI could erode as Gemini 3 gains users and outperforms OpenAI models on key benchmarks.​

What triggered the emergency

Google’s Gemini 3 launch is described as a powerful multimodal model that has impressed users and industry observers, with strong results in reasoning, coding, and other benchmark tests. Reports say Gemini 3 has quickly attracted large numbers of users and is deeply integrated across Google products such as Search, Android, YouTube and Workspace, increasing competitive pressure on OpenAI.​​

Altman is said to have warned staff that Gemini 3’s rise could create “temporary economic headwinds” for OpenAI and make external sentiment “rough for a bit” as the company fights to maintain its position. Internal memos reportedly frame this as a critical period for ChatGPT, with the company shifting into a “wartime footing” to respond.​

OpenAI’s internal response

According to multiple reports, Altman has raised OpenAI’s internal alert level to its highest tier, often referred to as “code red”, and ordered teams to put “all resources on quality” for ChatGPT. Staff working on areas like infrastructure, research and product are being redirected toward making ChatGPT faster, more capable and more engaging to close the gap with Gemini 3.​​

Employees directly responsible for ChatGPT are reportedly required to attend frequent or even daily check‑ins to track progress on improvements. The company is also said to be preparing new reasoning‑focused models that leadership claims will surpass Gemini 3, and some reports mention upcoming releases (such as updated GPT‑5.x variants) aimed at regaining the performance lead.​​

Projects paused or delayed

To free engineering capacity, OpenAI has reportedly paused or slowed several initiatives not seen as core to immediate ChatGPT competitiveness. These include:​

• Advertising and broader monetisation features inside ChatGPT.​​
• AI shopping and health agents and other consumer assistants.​
• A planned personal assistant product known as Pulse and some related integrations.​

Some commentary notes that delaying these revenue‑generating projects increases financial risk for OpenAI, since the company must still fund the high computing costs of running large models while putting growth experiments on hold.​

Wider industry implications

Commentators describe this as a role reversal from late 2022, when Google declared its own internal emergency after ChatGPT’s launch; now OpenAI is under pressure from Google’s improved models and massive infrastructure advantage. Analysts point out that Google can more easily offer advanced models at low or no direct cost because of its broader business, while OpenAI relies heavily on monetising ChatGPT and its APIs.​

The situation is framed as part of a broader “AI arms race” in which OpenAI, Google and other players like Anthropic and xAI are rapidly iterating on model quality, safety and cost. Altman has reportedly tried to reassure staff that, despite the emergency, he would not trade positions with any other company and believes OpenAI can still regain the initiative with its next generation of models.