OpenAI has introduced the new GPT-5.5 model, highlighting that this version is smarter, more capable, and the most natural the company has developed so far. It excels in coding, processing, and deep research, even with fewer instructions or guidance. The standout feature is its increased ability to "take action independently," working across multiple tools to complete tasks fully.
Greg Brockman, President of OpenAI, said that what makes this model special is its enhanced capability to perform tasks even with minimal instructions.
"It can analyze unclear problems on its own and determine the next steps to take. To me, it feels like laying a new foundation for how we will use computers and perform work through them in the future," he said in an interview with the media.
The launch of GPT-5.5 came less than two months after the release of GPT-5.4, reflecting a faster transition and development pace by AI developers amid intensifying competition, especially in the field of Agentic AI.
OpenAI is rushing to advance technology to compete with major rivals like Google and Anthropic, whose latest model, Claude Mythos Preview, is attracting significant interest from Wall Street investors.
Earlier in March, OpenAI closed a fundraising round, securing $122 billion, exceeding its $110 billion target, with investors including SoftBank, Amazon, Nvidia, Microsoft, and Andreessen Horowitz participating.
Following this funding round, OpenAI CEO Sam Altman faces increased pressure to justify the company’s soaring valuation, particularly as preparations are underway for a future initial public offering (IPO).
In recent months, OpenAI has noticeably cut expenses, scaling back some large investment plans and shutting down certain features and products like Sora, a short video app, to control costs and manage resources more efficiently.
In March, OpenAI reported monthly revenues exceeding $2 billion and total revenues of $13.1 billion the previous year, but the company still has not turned a profit despite massive investments.
According to OpenAI, "GPT-5.5 is the smartest and most natural-to-use model the company has developed, marking a significant step toward a new way of working with computers in the future."
GPT-5.5 quickly understands user intent and can independently perform more tasks. It excels in many areas, including writing and debugging code, online research, data analysis, document and spreadsheet creation, software control, and working across multiple tools to complete tasks.
Instead of users controlling every step in detail, they can assign complex, multi-part, or ambiguous tasks to GPT-5.5. The model plans, uses tools, self-checks, manages uncertainty, and continues working until completion.
GPT-5.5’s advancements are especially notable in Agentic Coding, knowledge work on computers, and early-stage scientific research—fields requiring multi-contextual thinking and continuous action.
Importantly, despite its significantly increased capabilities, GPT-5.5 maintains high processing speed, with per-token latency close to GPT-5.4 in real use, even while being noticeably smarter. It also uses significantly fewer tokens to accomplish the same tasks on Codex.
Regarding safety, OpenAI states that GPT-5.5 includes the strictest safeguards ever implemented to prevent misuse, while still enabling creative and beneficial uses. The model was evaluated through the organization’s full safety and readiness framework, with additional specialized testing for advanced capabilities such as cybersecurity and biology, and feedback was gathered from nearly 200 partner users before launch.
GPT-5.5 has been gradually made available to paying users—including Plus, Pro, Business, and Enterprise subscribers—via ChatGPT and the coding assistant Codex since last Thursday. API access will follow soon, with the company noting that different safety measures are required for such usage.
Read more
Follow the Facebook page: Thairath Money at this link - https://www.facebook.com/ThairathMoney