OpenAI on Thursday introduced its most superior synthetic intelligence mannequin, GPT-5.2, and mentioned it is one of the best providing but for on a regular basis skilled use.
The mannequin is best than predecessors at creating spreadsheets, constructing shows, perceiving photographs, writing code and understanding lengthy context, OpenAI mentioned. It will likely be out there beginning Thursday inside OpenAI’s ChatGPT chatbot and its utility programming interface (API).
The announcement comes weeks after OpenAI launched its GPT-5.1 mannequin. Rivals Anthropic and Google additionally launched new fashions final month, prompting OpenAI to declare a “code purple” effort to enhance ChatGPT and sideline different initiatives.
It is all a part of a high-stakes battle amongst main tech firms to create probably the most extensively used mannequin as shoppers and companies more and more add AI into their each day lives and workflow. OpenAI is relying on its GPT household of fashions to outline the long run as the corporate seeks to justify its $500 billion valuation and over $1.4 trillion in deliberate spending.
“We introduced this code purple to essentially sign to the corporate that we need to martial sources in a single explicit space, and that is a method to actually outline priorities and outline issues that may be deprioritized,” Fidji Simo, CEO of functions at OpenAI, informed reporters in a briefing on Thursday. “We have now had a rise in sources centered on ChatGPT on the whole, I might say that helps with the discharge of this mannequin, however that is not the explanation it is popping out this week particularly.”
OpenAI CEO Sam Altman informed CNBC on Thursday that Google’s launch of its Gemini 3 mannequin had much less of an impression on the corporate’s metrics than it initially feared. He mentioned he expects OpenAI will exit its code purple by January.
“I imagine that when a aggressive risk occurs, you need to give attention to it, take care of it shortly,” Altman mentioned.
OpenAI mentioned GPT-5.2 might be out there in Immediate, Considering and Professional variations. Immediate is quicker at writing and data looking for, Considering is best at structured work like coding and planning and Professional will ship probably the most correct solutions for troublesome questions, OpenAI mentioned.
The corporate mentioned that the mannequin tops trade benchmarks, together with SWE-Bench Professional, which evaluates agentic coding efficiency, and GPQA Diamond, a graduate-level scientific reasoning benchmark. On GDPval, an analysis OpenAI launched earlier this 12 months, GPT-5.2 beat or tied prime trade professionals on 70.9% of well-specified duties, the corporate mentioned.
“This has been within the works for a lot of, many months,” Simo mentioned. “Whereas we’re proud that we’re capable of have a cadence of releasing fashions quick, this explicit integration has been within the works for some time.”
Anthropic’s newest mannequin, Opus 4.5, scores greater than GPT-5.2 on SWE-Bench Verified, a check set that evaluates an AI system’s software program coding talents. OpenAI informed reporters that the benchmark is much less “contamination resistant, difficult, various and industrially related” than SWE-Bench Professional.
OpenAI was based as a analysis lab precisely 10 years in the past, but it surely’s turn out to be one of many fastest-growing business entities on the planet following the launch of ChatGPT in 2022. Greater than 800 million individuals now use its chatbot each week.
WATCH: The AI strain cooker: OpenAI, Nvidia and Google all on the transfer
Leave a Reply