OpenAI’s Hidden Aces: GPT-4.5 Exists, Top 50 Coding Model, AI Agent by Year-End
Explore OpenAI's latest advancements in GPT, multi-modal agents, and breakthrough AI research as revealed by Sam Altman in Tokyo amidst DeepSeek's global impact
"AI Disruption" publication New Year 30% discount link.
While DeepSeek is shaking up the global AI landscape with its low-cost models, OpenAI CEO Sam Altman is currently on his Tokyo tour.
Before finalizing a massive investment of $500 billion with SoftBank founder Masayoshi Son, Sam Altman, and OpenAI CPO Kevin Weil visited the University of Tokyo. During a Q&A session with faculty and students, they revealed some even more explosive news:
"Within OpenAI, we've already reached GPT-4.5, and reaching GPT-5.5 does not require more than 100× the computing power."
"Advances in inference models and reinforcement learning techniques have dramatically improved computational efficiency — allowing smaller models to achieve GPT-6–level performance without needing over 100× the compute."
"We will integrate all modalities together. You can see on the canvas: it will speak with you while simultaneously writing and compiling code for you. It will also be able to browse the internet."
"The 'o' model will support visual recognition capabilities. For example, if a piece of hardware needs maintenance, you can take a photo and the o-series model will be able to provide technical support."
"The o3-mini hints at our research direction for the next six to twelve months... I hope that by the end of this year, we can develop an intelligent agent model that can handle all high-difficulty tasks—with the exception of scientific discovery. It might need several hours to think, and it may even call upon a slew of tools, but in the end, it will be able to complete tasks for you."