One of the pioneers of the field of artificial intelligence OpenAI, long -awaited today GPT-4.5 Officially Introduced.
GPT-4.5 is available today as a research preview. The GPT-4.5, which OpenAI says ız Our biggest model so far, is not the company’s most talented or newest flagship model, unlike expectations. The new model, which may not be able to perform as well as O1 or O3-Mini on some issues, improves the calculation efficiency of the GPT-4 by 10 times. The “hallucination” status comes with a significant release of 4.5, which has been significantly reduced compared to GPT-4O, with better writing skills and is with a much wider general knowledge repertoire. First of all, “Pro” said, Sam Altman, OpenAI CEO, about the model opened to users: “This was the first model that made me feel like I was talking to a thoughtful person. There have been a few situations where I was amazed to get really good advice from an artificial intelligence. ” Altman said that the model was not designed to break the “benchmark” records.
You may be interested in
The company had recently taken another important step. OpenAI, introduced in recent weeks Deep Research He first opened his vehicle to the Chatgpt Pro subscribers. The company recently made an announcement, the system within certain limits of all Chatgpt Plus, Team, Edu and Enterprise users said. Developed deep research tool, Complex focuses on tasks and can conduct detailed research on the internet. To create a comprehensive report according to the commands given The system that finds, analyzes and synthesize hundreds of online sourcesIn the process, a special specially optimized for browsing and data analysis in the process.O3”Using the model. Using a large amount of text, painting and pdf search, interpretation and analysis processes Deep Research about the company, “… An important step towards our goal of developing AGI ..”.
In recent months, we had Gemini Deep Research. Google About Own Deep Research System had transferred: “Gemini Deep Research is your personal research assistant that helps you get detailed information on everything. It divides your request into a few relevant research questions, automatically analyzes sites on your behalf to access the relevant information, and creates comprehensive reports for pages on each subject. He does all this in a few minutes. Your reports are presented in a comprehensive viewer where you can edit with Gemini, and even open in Google Documents. Google One AI can be used in English in Gemini Advanced within the scope of the Premium plan. Gemini Advanced Model Open Menu “Try by selecting Gemini 1.5 Pro” with Deep Research. “
https://www.youtube.com/watch?v=jv-lpısnloo
Before that, the “operator / operator” system made a sound. Can perform transactions on websites instead of users “Operator / Operator” The system was launched for Chatgpt Pro users in the US last week. The web -based system, which is currently in the preview version, by the company “Needed gan assistant who can use his own browser to fulfill the folders”. When it is opened to everyone here The operator, which can be used from the connection, can analyze all items on a web page that it opens in its own browser through written commands, enter the text in case of need, click on the buttons or scroll on the page.
Operator, which reveals all the steps in real timeIn many subjects such as making users a food reservation or finding suitable hotels It reveals the potential to provide serious time earnings. The operator can think to fix himself and leave the control directly to the user if he is installed. If you want precise information such as logging in a website, the system, which wants to take over the control from the user, will be opened to all paid subscribers as part of the chatgpt in the future.