What is better than one Broadcast This can help you with tasks? One that can do for you. Openi continues to build its AI agents inside Chat GPT With the beginning of deep research.

Deep research

On Sunday, the Openi unveiled deep research, an AI agent who can do a multi -faceted research for you by pulling a strong amount of information from the web and combining these sources in a comprehensive report for you. Once pointed, deep research can operate completely freely. This is like having a research analyst on your command.

Strengthening deep research is a version of the Open AIO 3 model that is better than web browsing and data analysis. By taking advantage of O3’s advanced reasoning capabilities, it can find and interpret large content from the web, including text, images and more, and then output in a report that targets its needs Is

Each report is produced in five to 30 minutes in terms of work. However, you can work on other tasks during this time, improving the productivity of your workflow. The expired report is out in the chat. In the coming weeks, the agent will include photos, data ideas and more.

Nos: How General AI means better consumer experiences – see a bank’s view

According to the open, humans will take hours in the same task. In addition, the agent aims to be particularly good at finding niche information, which requires a number of searching humans.

According to Open, the target audience for deep research includes those who work very knowledgeable in finance, science, policy, and engineering. And those who need reliable, thorough research. Each report includes clear references and agents’ thinking so that users can double the information for themselves.

Double checking the chat boot response is usually good practice, as chat boats are deceitful. In particular, the open research warns that deep research can sometimes deceive the facts in response or make false defects, though, according to internal diagnosis, especially at a lower rate compared to the current chat GPT model “” Open also added that the agent can struggle to distinguish authentic information from rumors and, while highlighting the need for human reviews, fails to properly disclose uncertainty.

Comparison of performance

I Blog Post Announcing this feature, the GPT -4O versus deep research in the open includes the same results to show how the same indication produces very different results. People with deep research were very strong and better organized.

Deep research performed better to GPT-4O Last test of humanityAI Benchmark Examination recently launched by Scale AI and Center for AI Safety (CAI), which examines various articles on expert level questions. Deep Research pushed 26.6 % accuracy, GPT -4O, Grook -2, Claude 3,5 Sonts, Gemini thinking, O1, and even O -3 -Mini High, which just a few days ago The score was, as he highlighted. By Openi CEO Sam Ultman.

Openi also published the results of deep research performance on a series of other diagnostics, including cowia, which is a public standard that reviews AI on real -world questions and a specialist in various fields of deep research. Internal diagnosis of tasks. In both, the effective results of the deep research came out, even the GAIA external leader was topped on the board.

How to access

Due to computing power to run a deep research feature, only Chat GPT Pro users can access it right now. Month 200-Maha’s subscription A better version involves 100 questions and other facilities such as Chat GPT and Surah unlimited access and access to access. OperatorThe feature of its AI agent that can carry out the basic tasks of the browser, such as reservations.

Chat GPT Plus and team users will have access to the next, followed by enterprise and then free users. The Open AI shares that it plans to release a faster, more cost -effective version of the feature run by this model, which is small but as effective.

Nos: How General AI means better consumer experiences – see a bank’s view

If you want access to this feature right now but don’t want to finish $ 200 per month, Google has a similar feature, also known as Deep Research, which is available to all of its Gemini advanced users Is Google One AI Premium The plan, which costs $ 20 per month.

Back in December, Altman replied even one x User Those who told Altman with “K” to “Gemini but better,” do a deep research feature like Gemini, which suggested that the newly released deep research feature is an Openi response to Google –

Last week, Microsoft also announced a feature that is also called worth more good reasoning. Think deep, Which allows users to take advantage of Openi’s O1 reasoning model Providing high quality reactions to complex gestures. However, unlike the deep research features of Gemini and Open, it does not have agent’s capabilities or access to the Internet. The biggest benefit is that the experience is completely free.