A month after its last ChatGPT update, OpenAI announced GPT-5.2 on Thursday, a little more than a week after Sam Altman declared a ‘Code Red’ amid intense competition from Google’s Gemini 3. “We’re introducing GPT-5.2, the most capable model series yet for professional knowledge work,” OpenAI said in its release note. “Overall, GPT-5.2 brings significant improvements in general intelligence, long-context understanding, agentic tool calls, and vision—making it better at executing complex, real-world tasks end-to-end than any previous model.” Added Open AI. While GPT-5.2 is the latest update to OpenAI, executives from the firm said that it would be wrong to see the latest update as a response to Gemini 3, as per Axios. Reports, earlier, speculated that GPT-5.2 could arrive this week, given the ‘Code Red’ situation. What’s new in GPT-5.2? OpenAI said that while heavy users of ChatGPT Enterprise say it saves them up to 10 hours a week, GPT-5.2 is intended to be even more efficient than the 5.1 model. In particular, GPT-5.2 will be better than its predecessor at creating spreadsheets, making presentations, writing code, understanding images, understanding long contexts, using tools and handling complex, multi-step tasks, Open AI said. OpenAI also highlighted GPT-5.2’s prowess in coding, including reviews from early testers who had time to experiment with the latest model. The reviews were glowing: “GPT-5.2 represents the biggest leap for GPT models in agentive coding since GPT-5 and is a SOTA coding model in its price range. The version bump undersells the leap in intelligence,” Jeff Wang, the CEO of Windsurf, was quoted as saying. GPT 5.2 benchmarks OpenAI also said that GPT-5.2 reached new highs in benchmark tests for thinking, with scores showing a marked improvement by the latest model over GPT-5.1 in abstract thinking. Benchmark GPT 5.2 Think GPT 5.1 Think GDPval (wins or ties) Knowledge tasks 70.9% 38.8% SWE-Bench Pro (public) Software engineering 55.6% 50.8% SWE bench Verified software engineering 80.0% 76.3% science 76.3% diamonds Science 76.3% 2% 2 questions 88.1% CharXiv Reasoning (w/ Python) Scientific Numerical Questions 88.7% 80.3% AIME 2025 (no tools) Competition Math 100.0% 94.0% FrontierMath (Level 1–3) Advanced Math 40.3% 31.0% Advanced 40.4% Math (Tier 4Math. 12.5% ARC-AGI-2 (Verified) Abstract Reasoning 52.9% 17.6% GPT-5.2 Instant, Thinking, and Pro will begin rolling out in ChatGPT itself on Thursday Competing with Google’s Gemini After declaring ‘Code Red’ in early December, an existential threat to his product amid increasing competition from Google’s Gemini, OpenAI CEO Sam Altman gave a surprising response to CNBC on Thursday, saying that the launch of Gemini 3 had less of an impact than he expected. “I believe that when a competitive threat occurs, you want to deal with it quickly,” Altman told CNBC, adding that he expects his company to exit the ‘Code Red’ situation by the end of January 2022 Pro can bring any idea to life with its modern reasoning and multimodal capabilities. It significantly outperforms 2.5 Pro on every major AI benchmark,” Google announced, adding that the launch of Gemini 3 was “another big step on the road to” artificial general intelligence (AGI).