Just before the Thanksgiving holiday, Google made headlines by declaring that its latest artificial intelligence system, Gemini 3, had surpassed the capabilities of its emerging competitor, OpenAI. The company confidently positioned Gemini 3 as the most advanced AI model available.

Yet less than a month later, OpenAI responded by unveiling GPT-5.2, a new generation model it described as its most capable version for practical, professional applications. In a company announcement, OpenAI emphasized that GPT-5.2 achieved top scores on industry benchmarks covering software development, mathematics, and scientific tasks.

For many analysts, however, the real story was not which company held the top spot—it was that the distance between OpenAI’s foundational model and those developed by its rivals had nearly disappeared. This increasingly competitive environment is emerging at a delicate moment for OpenAI, which is working to reconcile ambitious spending commitments with revenue that has not yet caught up.

OpenAI expects to generate monthly revenues by late 2025 that would total roughly $20 billion a year, according to CEO Sam Altman. Even so, the company remains far from profitability. Over the next several years, OpenAI plans to invest approximately $1.4 trillion in computing infrastructure to support the development and deployment of its AI platforms.

When ChatGPT launched at the end of 2022, it ignited the global AI boom and gave OpenAI a commanding lead. That advantage lasted more than two years. But in the past year, AI developers across the United States and China have created systems that now match—and in some cases exceed—OpenAI’s most advanced models.

“The broad framework for building foundational models is now well understood,” said Rayan Krishnan, CEO of Vals AI, which monitors AI performance. “Every major lab is following similar approaches.”

Competition has intensified quickly. One week after Google released its updated AI model, San Francisco–based Anthropic introduced Claude Opus 4.5, which performed on par with OpenAI’s newest system. On Thursday, New York AI lab Runway debuted a video generation model that surpassed OpenAI’s Sora in standard benchmark tests.

Still, OpenAI continues to refine its technology at full speed. GPT-5.2 showed notable gains in programming tasks and specialized fields such as health care and finance. The company has also increased pricing, charging customers roughly 40 percent more for access compared to earlier models. The release coincided with Disney’s announcement of a new investment in OpenAI and an agreement to license Disney characters for use in Sora.

Following Google’s unveiling of its enhanced chatbot in November, Altman issued a “code red” memo to employees, urging them to accelerate improvements to ChatGPT and pause work on certain longer-term projects, according to a person familiar with the communication. ChatGPT remains the company’s flagship asset, used weekly by more than 800 million people—representing a 76 percent global market share, based on Similarweb data.

“Consumer AI is essentially synonymous with OpenAI,” Krishnan noted. “Losing that position would drastically change the company’s value.”

Altman instructed teams to focus on three priorities: speeding up ChatGPT’s responses, reducing unnecessary refusals to answer questions, and offering greater personalization options. These goals, he said, must come before the company’s emerging initiatives in health care, advertising, and e-commerce. To accelerate the effort, OpenAI temporarily reassigned employees from other teams to support the intensive development sprint.

ChatGPT remains OpenAI’s primary revenue engine. Approximately 6 percent of its massive user base pays $20 per month for premium access. The company is also developing new business models to generate income from the free version of the chatbot.

For the coming weeks, OpenAI plans to concentrate on ChatGPT advancement. With GPT-5.2 launched, the company is preparing for a broader rollout of updates early next year.

Altman is known for sending motivational directives during pivotal moments. Earlier this year, he issued a similar call to action when Chinese AI firm DeepSeek impressed the global tech community with its chatbot system. OpenAI quickly improved its own models before returning to longer-term projects.

Among those projects are several aimed at monetizing the free ChatGPT experience. Although early OpenAI employees took pride in avoiding ad-driven business models, the company’s significant computing expenses have pushed it to explore new revenue streams. These include advertising experiments designed to be integrated subtly into chatbot interactions, as well as commerce-related features that allow users to shop directly within ChatGPT. If a user purchases a handcrafted ceramic vase from an Etsy seller, for instance, OpenAI would receive a share of the sale.

Despite the memo’s emphasis on consumer products, ChatGPT is only one side of OpenAI’s strategy. The company is also expanding rapidly into the enterprise software market, using the same foundational technology behind ChatGPT. Internally, OpenAI executives often describe the organization as two businesses—one focused on consumers and another focused on commercial clients. They believe AI will eventually reshape nearly every software category.

Over the last year, OpenAI has applied reinforcement learning techniques to strengthen model performance in fields such as mathematics and programming. Through repeated trial and error, models learn specialized skills and become more reliable. AI tools that generate software code have become major sources of revenue, with some developers and researchers paying up to $200 per month for access. GPT-5.2 is designed in part to improve these capabilities further.

The company is now using similar training strategies to develop models optimized for finance, health care, and other industries. This approach aims to expand OpenAI’s business-to-business product line.

“AI-based coding tools have already changed workflows in Silicon Valley,” said Wei-Lin Chiang, cofounder and CTO of LMArena, which evaluates AI systems. “And we should expect similar transformations in other sectors.”

According to new benchmark data from Vals AI, OpenAI’s latest models outperform competitors in tasks related to legal analysis and financial decision-making, signaling its push into specialized enterprise markets. The company is also building an enterprise software division modeled after those at tech giants like Google.

However, OpenAI faces challenges as it competes head-to-head with Google. Google’s enterprise customers already depend on tools such as Gmail and Google Docs, making it easier to bundle new AI services. To counter this, OpenAI introduced its own AI-optimized web browser this fall, marking a direct challenge to Google Chrome.

“This race will unfold across many different battlegrounds,” Krishnan said.