OpenAI on red alert as Gemini 3 and Claude Opus 4.5 threaten 'AI throne'

OpenAI CEO Sam Altman issued a 'code red', ordering the entire company to focus on improving ChatGPT as competitive pressure from Google Gemini 3 and Anthropic Claude Opus 4.5 increased at an unprecedented rate.

OpenAI CEO Sam Altman speaks to the media after a Q&A session at the OpenAI data center in Abilene, US, on September 23, 2024 - Photo: REUTERS

On December 1, Mr. Sam Altman sent a "red alert" internally to the company, asking to focus on improving ChatGPT in terms of response speed, reliability and personalization. OpenAI postponed many side projects such as advertising plans, developing automatic AI assistants in healthcare , shopping and personal assistant "Pulse".

The decision reflects a clear concern: despite more than 800 million weekly users, ChatGPT is no longer confident of maintaining its leading position if it does not quickly upgrade.

Google, Anthropic accelerate

The Guardian newspaper commented that the appearance of Gemini 3 developed by Google is a strong blow to ChatGPT's position. According to international reports, Gemini 3 passed many important benchmark tests with strong multi-modal reasoning and processing capabilities.

Meanwhile, according to Business Insider, not only regular users, but many business leaders also expressed their admiration for Gemini 3.

Notably, Mr. Marc Benioff - CEO of software giant Salesforce - declared "no return" to ChatGPT, evaluating Gemini 3's leap as "terrible" in terms of reasoning ability, speed and multi-modal processing ability.

The Guardian believes that Google is taking advantage of its cloud computing infrastructure, huge data and financial resources to develop an "all-in-one" AI model, which is something that smaller competitors will find difficult to catch up with.

Not to be outdone, Anthropic has just launched Claude Opus 4.5 - a model they claim is "the most powerful ever", with the goal of dominating the field of programming, AI agent applications and computer usage.

In the SWE-Bench Verified benchmark programming test, Opus 4.5 achieved 80.9% success - the highest among today's top AI models, surpassing both Gemini 3 Pro and the latest version of OpenAI.

Opus 4.5 is designed to handle tasks like automated programming, source code review, source code refactoring, workflow management - features that many companies are looking for. This shows: while Gemini 3 scores with "versatile, multi-modal", Claude Opus 4.5 goes for "technical depth" and can occupy the segment that ChatGPT is less focused on.

ChatGPT's position is shaky.

Reuters news agency assessed that the "weapons" that helped ChatGPT break through, such as speed, convenience, and good conversation ability, are no longer enough to retain users when competitors offer stronger experiences in terms of reasoning, multi-modal processing, source code, and business processes.

Gemini 3 and Claude Opus 4.5 "attack" in two directions: one is versatile, easy to use for everyone; the other is specialized, suitable for business and programming. As user needs become more and more diverse, the "comprehensive model" is a big advantage.

Sam Altman’s “red flag” shows that OpenAI understands the pressure: it’s impossible to develop new features while maintaining quality when competitors are hot on its heels. But it also makes it easy for users to see that ChatGPT has “issues that need fixing,” and many will try to switch to Gemini 3 or Opus 4.5.

Despite its large user base and reputation, as many analysts have pointed out, ChatGPT is facing a “fierce challenge” if it does not continuously improve. What was once an advantage can now become a “stagnant factor” if it does not innovate.

The competition also reveals one thing: there is no “default king” in the AI ​​world . Widespread popularity and strong brands are no longer enough to guarantee position if the product does not develop strongly and suit diverse needs.

If OpenAI truly improves ChatGPT—improving its speed, stability, personalization, and inference capabilities—it has a chance to hold on. But if it improves too slowly or fails to keep up with its rivals, the “old king” could be toppled.

The AI ​​race is not just about creating a "more powerful" model, but "who comes up with the product that best suits real needs, fastest, most efficient, most reliable". There is no safe position, only continuous development or being left behind.

The race is not for the weak

The AI ​​market is witnessing unprecedented fierce competition. OpenAI has 800 million users per week but has just issued a "red alert" urging reform. Google takes advantage of its huge infrastructure to launch Gemini 3 with superior multi-modal processing capabilities.

Anthropic took a different approach: focusing on the professional segment, with the Claude Opus 4.5 achieving an 80.9% success rate on the most difficult technical test. This strategic differentiation shows that there is no single “winning formula.” AI companies must constantly innovate and react quickly to the market, because being late means failure.

THANH HIEP

Source: https://tuoitre.vn/openai-bao-dong-do-khi-gemini-3-va-claude-opus-4-5-de-doa-ngoi-vuong-ai-20251203214209418.htm