OpenAI to Expand Access to Deep Research AI Agent Across ChatGPT Tiers

OpenAI Unveils Deep Research AI Agent

OpenAI has introduced a new AI agent called "Deep Research," powered by its upcoming o3 reasoning AI model. This autonomous assistant is designed to scour the web and scholarly sources for information, compiling comprehensive reports on user-specified topics 1

1

. Initially limited to ChatGPT Pro subscribers, OpenAI CEO Sam Altman has now announced plans to make Deep Research more widely available across different subscription tiers 2

2

.

Expanded Access and Usage Limits

Altman revealed that ChatGPT Plus users ($20 per month) will receive 10 uses of Deep Research per month, while free tier users will get 2 uses per month. The company intends to scale up these limits over time 1

1

. This move aims to make the powerful AI tool accessible to a broader audience while potentially encouraging users to upgrade to higher-tier subscriptions.

Capabilities and Potential Impact

Deep Research has been praised for its ability to accomplish in minutes what would take humans hours to complete. The AI agent can find, analyze, and synthesize information from hundreds of online sources to create comprehensive reports at the level of a research analyst 2

2

. Some users have described it as "utterly transformative" in how they engage with information 1

1

.

Technical Insights and Challenges

Deep Research employs a "test-time" computing method to produce detailed reports, which can take anywhere from 30 minutes to several hours 3

3

. The model uses a technique called "test-time thinking," which forces it to fact-check itself before answering questions by breaking queries down into multiple related sub-questions.

Future Developments and Model Unification

Altman also shared OpenAI's plans to consolidate its various AI models. The company aims to create a "magic unified intelligence" that can use all of OpenAI's tools and determine when to engage in extended reasoning 3

3

. This includes the upcoming GPT-4.5 model (internally called Orion) and the integration of o-series and GPT-series models.

Accuracy Concerns and Industry Challenges

Despite the excitement surrounding Deep Research, concerns about the accuracy of AI language models persist. Recent incidents involving incorrect information generated by AI chatbots highlight the ongoing challenges in ensuring reliable outputs 3

3

. As AI tools become more widely available, addressing these accuracy issues remains a critical priority for the industry.