Openai, a familiar name in the world of artificial ielligence, has just unveiled a new ChatGpt feature called Deep Research. This tool operates in such a way that it can automatically plan and execute the process of finding the information they need. According to Openai, Deep Research is able to “plan a multi -step route to obtain your desired information and, if necessary, rely on instaaneous data, modify and respond.” This flexibility and adaptation to variable conditions are key features of this new tool.
The Deep Research feature is not limited to the production of text. During the research, this smart age shows a summary of its work process along with a list of resources and a brief description of the process in the sidebar. This transparency in performance allows users to know how the system is reached and more confide in the results. Users can ask questions using a variety of inputs such as text, images and side files such as PDF or spreadsheets, such as text, images and accessories. The process by Deep Research may take between 1 and 2 minutes, and Openai has announced that in the future, it will be possible to add images and embedded charts to the answers.
Despite the remarkable abilities, OpenAI has also cited limitations for Deep Research. According to the company, the tool “sometimes gets illusion” and may make information. This shows that the system faces challenges in recognizing the difference between valid and invalid information, as well as in determining the amou of confidence in the answers.
However, the developme of productive artificial ielligence tools that go beyond the production of simple text is the future that companies like Openai have drawn for ages. OpenAI claims that Deep Research can perform a performance at the level of an expert researcher, and this is an importa step towards realizing this landscape. In a show preseed by Deep Research, the system was applied with information on the change of retail industry in the last three years, and the response provided included key pois and related tables. This example illustrates the system’s ability to analyze data and prese organized results.
The DEEP Research launch comes shortly after the iroduction of the Operator tool. Operator is a tool that allows users to use the web browser to perform differe tasks on their computer. It should be noted that a few mohs ago, Google unveiled the same name for Gemini artificial ielligence that does so.
Openai has allowed users to pay up to 2 questions per moh for users who pay a $ 4 mohly fee. Also, for Plus users, team and finally corporate users, “limited access” has been promised.
Openai calls Deep Research “a very heavy computing capability” and pois out that the longer the research time, the need for inference calculations increases. The ChatGPT makers have announced that in the future, by providing faster and more affordable versions, they will have less restrictions on all money users.
Openai has proudly announced that the model that feeds Deep Research has set a new record in the Humanity’s Last Exam Benchmark. This test assesses the ability to answer specialized questions at high -level questions. The Deep Research model, utilizing browser and Python tools, has achieved a accuracy of 4.9 %, which is far more than the GPT-4O’s 2.5 % accuracy. The O3-Mini (High) model, which was solely evaluated on the basis of the text, finished second with 2 % accuracy. These achievemes indicate Openai’s significa progress in developing artificial ielligence and its research capabilities, and draws a promising future for this technology.
25,199,000
23,899,000
Toman
Source: Android Authority
