OpenAI Unveils GPT-5.2 Amidst ‘Code Red’ Challenges

OpenAI has unveiled GPT-5.2, its most advanced artificial intelligence model yet, showcasing enhanced capabilities in writing, coding, and reasoning benchmarks. This rollout follows just a few days after CEO Sam Altman declared a “code red” internally, initiating a company-wide effort to enhance ChatGPT in response to escalating competition.

“We announced this code red to clearly indicate to the organization our intent to concentrate resources in a specific area, effectively establishing our priorities,” stated OpenAI’s CEO of applications, Fidji Simo, during a press briefing on Thursday. “We’ve seen an increase in resources directed towards ChatGPT overall.”

Simo refuted claims that the launch of GPT-5.2 was expedited due to the code red, asserting that the model’s release has been in development for several months. Nonetheless, she noted that the additional resources focused on ChatGPT have proven to be beneficial.

While OpenAI’s models and products were regarded as top-tier when ChatGPT debuted in 2022, this status is no longer guaranteed. The company is now facing a variety of formidable competitors, notably Google, whose recently released Gemini 3 model has received positive feedback from the tech community. Google’s Gemini app has experienced significant growth over the past year, boasting over 650 million monthly active users compared to OpenAI’s 800 million weekly active users. This competitive pressure has compelled OpenAI to scale back some of its more ambitious initiatives, including the consideration of ads in ChatGPT, and to concentrate on enhancing its core technologies and products.

Similar to OpenAI’s recent model releases, GPT-5.2 is being launched in multiple iterations: Instant, designed for quicker responses and effective information retrieval; Thinking, which specializes in coding, mathematics, and planning; and Pro, the highest tier of OpenAI’s models that provides superior accuracy for challenging queries.

OpenAI labels GPT-5.2 as its finest model yet for everyday professional applications. GPT-5.2 Thinking achieved record scores on GDPval, an OpenAI-created benchmark that evaluates performance across 44 real-world professions between AI models and human experts. The company reports that the model outperformed human professionals in over 70 percent of tasks, completing them 11 times more swiftly.

Max Schwarzer, OpenAI’s post-training lead, indicates that the new release should significantly decrease occurrences of hallucinations. The company claims that GPT-5.2 Thinking exhibited 38 percent fewer hallucinations than GPT-5.1 on benchmarks assessing answers to factual inquiries.

OpenAI is making GPT-5.2 accessible to both ChatGPT users and developers through OpenAI’s API product. The company asserts that the new model series “delivers tangible improvements across both everyday and complex use cases.”