OpenAI has introduced its launching GPT-5.2 efficient instantly, with entry rolling out in phases beginning Dec. 11.
That is the second main replace to GPT-5, the long-awaited OpenAI mannequin that launched on Aug. 7.
In a weblog submit saying the brand new collection of ChatGPT fashions, the corporate known as GPT-5.2 the “most succesful mannequin collection but for skilled data work.”
This Tweet is at the moment unavailable. It may be loading or has been eliminated.
GPT-5.2 is definitely a collection of fashions; the collection itself is a part of the bigger GPT-5 collection. Here is what we’re taking a look at:
-
GPT-5.2 Immediate
-
GPT-5.2 Considering
-
GPT-5.2 Professional
When are you able to begin utilizing GPT-5.2?
With a Dec. 11 launch date, GPT-5.2 is accessible now — although not all customers will see it listed as an out there mannequin in ChatGPT.
Mashable Mild Pace
As with earlier mannequin rollouts, OpenAI will first deploy the mannequin to paid customers on Plus, Professional, Go, Enterprise, and Enterprise plans.
Even for paying prospects, OpenAI tends to make its new fashions out there in phases — so if it is not out there instantly through the OpenAI API or ChatGPT, preserve checking.
“We deploy GPT‑5.2 progressively to maintain ChatGPT as easy and dependable as we are able to,” the corporate wrote. “Should you don’t see it at first, please strive once more later. In ChatGPT, GPT‑5.1 will nonetheless be out there to paid customers for 3 months underneath legacy fashions, after which we’ll sundown GPT‑5.1.”
You probably have a paid ChatGPT account, go to ChatGPT.com to see if the mannequin is accessible. Ultimately, all ChatGPT customers ought to have free entry to GPT-5.2, although that entry will probably be restricted on free membership plans.
GPT-5.2 benchmarks and efficiency
OpenAI printed a new system card that exhibits GPT-5.2 makes iterative enhancements on key benchmarks in comparison with GPT-5 and 5.1. As well as, OpenAI stated that GPT-5.2 hallucinates lower than earlier fashions.
-
GDPval: GPT-5.2 scored 70.9 %, whereas GPT-5 scored 38.8 %
-
SWE-Bench Professional: GPT-5.2 scored 55.6 %, whereas GPT-5 scored 50.8 %
-
AIME 2025 math: GPT-5.2 scored one hundred pc, whereas GPT-5 scored 94 %
-
FrontierMath (Tier 1–3): GPT-5.2 scored 40.3 %, whereas GPT-5 scored 31.0 %
Disclosure: Ziff Davis, Mashable’s father or mother firm, in April filed a lawsuit in opposition to OpenAI, alleging it infringed Ziff Davis copyrights in coaching and working its AI programs.
[/gpt3]
