OpenAI introduced GPT-5, the company’s smartest AI model, has shown performance improvements across benchmarks for writing, programming, and reasoning. It comes only days after CEO Sam Altman declared internally that he was launching the product. “code red,” A company-wide effort to improve ChatGPT in the face of intense rivalry.
“We announced this code red to really signal to the company that we want to marshall resources in one particular area, and that’s a way to really define priorities,” Fidji SIMO, OpenAI’s director of applications, told reporters during a press conference on Thursday. “We have had an increase in resources focused on ChatGPT in general.”
Simo denies that OpenAI has accelerated GPT-5.2 due to its code red. She claims the company’s been working on this release for several months. She said that the resources devoted to ChatGPT were increased. “helpful.”
OpenAI was considered the best at what it did when ChatGPT first launched in 2022. But that is no longer true. Google’s recently released Google Assistant is perhaps the biggest threat to OpenAI. Gemini 3 model Tech industry has been very positive about the app. Google’s Gemini has seen a rapid growth in the past year. It now boasts more than 650 millions monthly active users versus OpenAI’s 880 million weekly active user. This pressure forced OpenAI’s most ambitious projects to be scaled back, such as its plans for introducing a new AI system. ads to ChatGPT, It will refocus its efforts on the improvement of core products and technologies.
GPT-5.2 will be available as a series of models, similar to the recent launches of the OpenAI models. These include Instant and Thinking. The former excels in coding, planning, and math. Pro is the strongest tier, with higher accuracy for difficult questions.
OpenAI has named GPT-5.2 as its most advanced model for professional everyday use. GPT 5.2 Thinking scored highest on OpenAI’s benchmark GDPval which measures the performance of AI models against human professionals in 44 different occupations. According to the company, GPT-5.2 Thinking outperformed human professionals on 70 percent of tasks and performed them 11 times faster.
Max Schwarzer is OpenAI’s Post-Training Lead. He says GPT-5.2 could also improve ChatGPT’s hallucinations. GPT5.2 Thinking, according to OpenAI’s post-training lead Max Schwarzer, hallucinated at a 38 percent lower rate than GPT5.1 in benchmarks measuring responses to factual question.
OpenAI is now bringing GPT-5.2 models to ChatGPT and API users. OpenAI claims that the new models are available. “brings clear gains across everyday and advanced use cases.”
