Gemini 3 launch had less of an impact on ChatGPT than CEO Sam Altman feared — OpenAI eyes January exit from “code red”

by

Like what you’re reading? To stay informed about the newest news, reviews, and features, consider setting Windows Central as a preferred source in Google Search. Learn more about the benefits of doing so to always be in the know.

OpenAI released its much-anticipated GPT-5.2 model on December 11, 2025, after Google’s Gemini 3 significantly outperformed ChatGPT in testing. CEO Sam Altman responded quickly, initiating a high-priority effort to address the competitive challenge.

With more people using Google’s Gemini, OpenAI might be preparing to launch GPT-5.2 to boost ChatGPT’s performance. This new AI model is expected to be much better at tasks like coding, creating videos, and working with images. It will be available to use directly within ChatGPT, or for developers through OpenAI’s API.

ChatGPT-5.2 ships in three different flavors: Instant, Thinking, and Pro versions. For context, the instant version is designed for writing and research, while the Thinking version takes on complex tasks like coding. As you might have guessed, the Pro version is designed to find responses to difficult questions.

The model is performing very well and has achieved excellent results on several important tests. This includes SWE-Bench Pro, which assesses how well it can act as an independent coder, and GPQA Diamond, a benchmark for evaluating its reasoning skills. It achieved top scores on both.

We issued a ‘code red’ alert to clearly communicate the need to focus all available resources on a specific priority. This helps us define what’s most important and what can be put on hold. While we’ve increased resources for ChatGPT overall, that’s not the primary reason for this week’s model release.

OpenAI CEO of apps, Fidji Simo

Interestingly, OpenAI CEO Sam Altman shared that the launch of Google’s Gemini 3 didn’t affect ChatGPT’s performance as much as some expected (according to CNBC). He also stated that OpenAI expects to resolve its recent internal “code red” situation by January. Altman explained, “When a competitor emerges, it’s best to address the challenge directly and efficiently.”

I’ve been seeing a lot of talk lately about AI potentially taking over jobs, and it’s getting a little scary! The latest version, GPT-5.2, is really adding fuel to that fire. Apparently, it absolutely crushed it on a test called GDPval, scoring a 70.9% on tasks where the instructions were really clear – even better than the professionals taking the same test! It makes you wonder if AI will really start showing up in workplaces soon.

Bill Gates, the co-founder of Microsoft, believes AI will eventually be able to do most jobs currently done by humans. However, he thinks people will still have control over deciding what uniquely human activities to keep. He playfully suggested that watching computers play baseball just wouldn’t be enjoyable.

Read More

2025-12-12 16:11