OpenAI Upgrades Deep Research with GPT-5.2, Adds Targeted Web Search

OpenAI Upgrades Deep Research with GPT-5.2: Websites Can Now Be Scanned

AI giant OpenAI announced that it has enhanced 'Deep Research,' one of the popular features of the ChatGPT platform, with the new GPT-5.2 model. This update now allows users to conduct searches by targeting specific websites and enables real-time data tracking. The development is considered a significant leap in AI-assisted research capabilities and appears poised to accelerate competition in the sector.

New Capabilities with GPT-5.2

The GPT-5.2 model is described as an integrated system that brings together OpenAI's various technologies. This system, which combines the long chain-of-thought and traditional model approaches the company previously worked on, aims to provide deeper and more contextual understanding, especially in research-focused tasks. The new Deep Research feature allows users to conduct searches by targeting a specific website, scan site content, and access up-to-date data.

This capability is reminiscent of some features of the 'Operator' AI agent previously introduced by OpenAI. Operator could perform human-like interactions such as navigating web pages, filling out forms, clicking, and scrolling. Similarly, GPT-5.2's Deep Research feature carries the potential to automate web-based research tasks.

Technological Infrastructure and Integration

One of the important points emphasized by OpenAI when announcing GPT-5.2 is that this model will now be presented not as a standalone model, but as a system integrating many of the company's technologies (such as o3). This indicates that OpenAI is experiencing a strategic shift from a pure GPT model development approach toward more integrated systems.

Programming-focused tools previously released by the company, such as Codex, are also seen as part of this integration strategy. Codex was a cloud-based AI agent designed to assist in software engineering tasks. GPT-5.2 aims to combine the capabilities of such specialized, task-oriented tools within a general-purpose research and production platform.

Competition in the Sector and Google's Response

OpenAI's move has further intensified competition in the AI-assisted search and research market. According to news sources, Google has also begun offering developers a new Deep Research agent powered by its Gemini 3 Pro model. It is claimed that Google's new agent has outperformed its competitors in the tests it has entered.

This competition between the two technology giants is an indicator that AI-assisted research tools will rapidly evolve and that users will be able to access more sophisticated, reliable, and in-depth research capabilities. However, this rapid development also brings with it some concerns.

Reliability and Accuracy Concerns

While experts appreciate the opportunities offered by AI-assisted deep research tools, they remain cautious, particularly regarding web content scanning and real-time data usage. How these tools will assess the accuracy and impartiality of information on the websites they scan, and how they will filter out outdated or misleading information, remain significant question marks.

Furthermore, it is noted that models like o3 have not yet achieved full autonomy but can accelerate the 'smart evolution of smart progress' by assisting researchers in areas such as data processing, model design, and code generation. This also points to the early stages of AI systems' capacity for self-improvement.

Future Expectations and Conclusion

The Deep Research feature enhanced by OpenAI with GPT-5.2 has the potential to be a groundbreaking tool for students, academics, journalists, and professional researchers. By targeting specific websites for in-depth analysis, it promises to significantly streamline the information gathering and verification process.