OpenAI drops Deep Research access to Plus users, heating up AI agent wars with DeepSeek and Claude

Open Announced that today is spreading strongly Deep research Allowing everyone Chatgpt Plus, Team, Education and Venture Users have significantly expanding the company’s most transformative AI agent from the company, since the original Chatgpt.

According to an announcement of Openai An account of official XMoreover, the team, education and enterprise users will first receive 10 deep research requests in the month, and Pro Tier subscribers will receive a monthly 120 survey.

Deep research equipped with a special version of Openai’s approximation O3 modelThe EU represents a significant turn that can help in complex research tasks. Unlike traditional conversations that respond immediately, deep research examines hundreds of online sources independently, analyzes text, pictures and PDFs and synthesizes comprehensive reports produced by professional analysts.

Deep research is now spread to all Chatgpt Plus, team, edu and enterprise users?

– Openai (oopenai) 25 February 2025

AI Research Weapons Race: Deepseek’s open problem meets Openai’s award game

Openai’s expanding rollout’s time is difficult random. The generative AI scene has become dramatically with China in recent weeks Depth emerges as an unexpected disruption. With open source DeepSEEK-R1 model each other I am my licenseThe company mainly objected to the closed, subscription-based work model, which determined the development of the West AI.

What makes this competition is particularly interesting is different philosophies in the game. Openai continues the most powerful opportunities behind more complexity Subscription stationsDeepseleek has chosen for a different approach to rooting: Give technology and bloom a thousand applications.

When DeepSeek announces waves, the open-source justification model announced when DeepSseek announces waves, the open-source justification model announced when he announces the waves, the EI announced.

However, for those who closely follow the AI ​​developments, DeepSEEK and R1 did not come out … pic.twitter.com/fuahyp0hhz

– Y Combinator (@ycombinator) 5 February 2025

This strategy is the technology setting in the beginning of the adoption of technology created more than indoor platforms on open platforms. Linux offers a parallel with an attractive history, dominated by the dominated in the server infrastructure. The enterprise has invested in property solutions that may include open alternatives that may include questions, emergency competition advantages, or to develop more updates in organizations.

Confusion Final integration Deepseek-R1 is in a part of Openai’s price point – this open approach demonstrates how quickly this open approach can give the competitive products. This time anthropic Claude 3.7 Sonnet He caught another way to transparency in the grounding process with “visible extended thinking”.

Deepseek’s R1 is a matter of how impressive model, especially for the price.

Obviously, we will provide better models and at the same time it is a legal invigorator to have a new opponent! We will draw some releases.

– Sam Altman (@sama) 28 January 2025

The result is a fragmented market that each great player now offers a different approach to research that works in AI. For businesses, this means more choice, but also has difficulty determining which platform to best match with their unique needs and values.

Wally Garden from the Public Square: The calculated democratic pivot of Openai

Sam Altman writes this deep research “Probably worth $ 1000 per month for some users“He only discovers more than price elasticity – acknowledges the extraordinary impairment between potential users. This admission is cut into the center of Openai’s ongoing strategic balance sheet.

The company faces a basic tension: “Artificial general intelligence protects the premium exclusiveness of the development of its development when performing the mission to ensure that all of mankind will benefit. Today’s announcement is a careful step to more accessible without breaking the income model.

I think it will first offer 10 use per month for Chatgpt Plus, and this time we will use it 2 months per month with the intention of scaling them.

Probably worth $ 1000 per month for some users, but I’m glad to see everyone doing with him! https://t.co/ybicvzodpf

– Sam Altman (@sama) 12 February 2025

Openai is essentially a tized offers the free Pier users, by limiting the two questions in the month – it is enough to demonstrate the capabilities of technology without cannibbalization. This approach is a classical “Freemium” playbook, which determines most of the digital economy, but every deep research is followed by unusual intensive restrictions that reflect the important sources required for survey.

Distribution Plus 10 months inquiry for users ($ 20 / month) Pro users create a clear degree that protects the premium value supply ($ 200 per month). This step-rolled strategy, Openai’s advanced AI opportunities demanding democratization, reducing price barriers and accepts how these opportunities are packaged and the main display of delivery.

Outside the surface: the hidden strong and surprising weaknesses of deep research

Title figure – 26.6% accuracy “The final examination of humanity“- This benchmark, designed to be extraordinary for human professionals, represents a quantum leap out of previous AI opportunities. For the context, this test would be significant a year before 10%.

What is the most important thing, not just raw performance, but the test itself is the test of the test, which applies nuanced justification outside the nature, different domains and pattern matches. Deep research combines several technological advances: Many steady planning, adaptation information and perhaps the most important is to recognize and eliminate their restrictions during the research process.

Again, these opportunities come with remarkable blind spots. The system remains sensitive to what can be called “consensus bias“- In spite of the prospect of the previous consideration of the protective view of the controlled point of view, it may be especially problematic in areas where the innovation is often made of wisdom.

Moreover, the power of transmission of the system to the existing website is to inherit the bias and restrictions of the source material. With the rapidly developing areas or limited online documentation, deep studies in Niche specialists can really fight to analyze thoroughly. And without access to property base or subscription-based academic magazines, certain specialized domains can remain inferior, despite the ability of fine thinking.

Executive toothpaste: Deep research reconciliates the rules of knowledge

For C-Suite leaders, the deep research presents a paradox: is a strong enough tool to redefine roles in organizations, but still very limited without careful human control. Immediate productivity earnings are undeniable – once required assignments required by analytical time days now can be completed within a few minutes. However, this efficiency comes with complex strategic results.

Organizations that effectively connect deep research, likely to reduce the information in the information. On the contrary, instead of replacing small analysts, the technology can create new hybrid roles, evaluating the framework questions, sources of frameworks, and the concepts of AI. The most successful applications will probably not look into deep research in deep research, not substitutive research for human judgment, but as an amplifier of human resources.

Deep research for CHATGPT PLUS users!

It’s one of my favorite things we have ever sent.

– Sam Altman (@sama) 25 February 2025

The assessment structure creates its strategic considerations. For Pro users with 120 surveys, each survey costs about $ 1.67 effectively – a meaningless expense compared to human labor. However, the questions of limited organizations create artificial scarce forcing the questions to prioritize the possibility of indeed-in-depth research. This restriction may lead to a more thoughtful application of technology, which is more encouraged to be more unlimited.

Long-term effects are deeper. Research opportunities limited to elite organizations will increase and competitive advantage will increase, but to learn how organizations resolve the questions and decision-making processes. From the knowledge of the concept of strategic value – it varies from collection to the generation of concept.

Message for technical leaders is clear: AI research revolution is no longer coming – here. The question is not to match, but it is not possible to develop the processes needed to develop in a landscaped landscaping of deep research, however, the need to develop skills and cultural thinking.