OpenAI’s Deep Research Evolution: New Full-Screen Viewer and Site-Specific Controls

In the ever-evolving landscape of artificial intelligence, OpenAI has just taken a massive leap forward for power users and researchers alike. As of February 2026, the company has officially rolled out a transformative update to its "Deep Research" tool—an agentic research assistant that first captured the tech world's imagination exactly one year ago.

This latest version is no mere cosmetic tweak. It introduces a high-fidelity full-screen viewer, granular source controls, and a more capable backbone in the form of the GPT-5.2 model. For anyone who has ever felt limited by a narrow chat window while trying to analyze complex data, this update is a game-changer.

The Evolution of Deep Research

When OpenAI first launched Deep Research, it promised to do the "heavy lifting" of the internet: scouring hundreds of websites, synthesizing findings, and compiling expert-level reports in a fraction of the time it would take a human. However, early users often felt like they were watching a "black box" operation—you provided a prompt and waited for a final result, with little room to steer the ship mid-voyage.

The 2026 update directly addresses these pain points. By integrating tighter controls and a dedicated document workspace, OpenAI is transitioning ChatGPT from a conversational chatbot into a professional-grade research station.

A New Perspective: The Full-Screen Viewer

The centerpiece of this update is the built-in full-screen viewer. Previously, long-form reports were confined to the standard scrolling chat history, which made navigating 20-page analyses a chore.

Now, when a report is generated, it opens in a separate, dedicated window. This interface is designed for deep work:

• Interactive Table of Contents: On the left side of the screen, a dynamic index allows you to jump to specific sections—such as "Market Trends" or "Competitive Analysis"—with a single click.

• Source Sidebar: On the right, a persistent list of sources provides "receipts" for every claim made. This transparency is crucial for professionals who need to verify the primary data before presenting it to stakeholders.

• Focused Reading: By separating the report from the chat history, users can focus on the content without the distraction of previous prompts or unrelated side conversations.

Precision Research: Focusing on What Matters

Perhaps the most requested feature since the tool's inception has been the ability to tell the AI where to look. In this update, OpenAI has introduced the "Manage Sites" feature.

Instead of letting ChatGPT scour the open web at random, users can now:

1. Specify Trusted Domains: You can restrict the research to specific domains, such as .gov sites, academic journals like JSTOR, or industry-specific news outlets.

2. Connected Apps: For the first time, Deep Research can tap into your connected productivity apps. If you have granted permissions, the GPT can pull insights from your own Google Drive, SharePoint, or GitHub repositories, blending public web data with your private organizational knowledge.

3. Real-Time Steering: You no longer have to wait for the report to finish before realizing the AI took a wrong turn. You can track progress in real-time and "edit the scope" or "add new sources" while the chatbot is still generating.

The Power of GPT-5.2

Underpinning these features is the newly updated GPT-5.2 Instant model. While previous versions focused on raw reasoning, the 5.2 iteration is tuned for "measured and grounded" output. This means fewer hallucinations and a more professional, contextually appropriate tone for business and academic reports.

For technical users, this model provides clearer output for how-to guides and complex advice. Whether you are drafting a 1,500-word blog post or a technical white paper, the GPT now handles nuance with a level of sophistication that feels less like a machine and more like a senior analyst.

Seamless Export and Integration

In professional environments, a report is only as good as its shareability. Recognizing that users need these findings in their own workflows, the new viewer supports instant downloads in several formats:

• Markdown: Ideal for developers and bloggers who want to port the content directly into CMS platforms.

• Word (.docx): The gold standard for business editing and collaborative reviews.

• PDF: Perfect for high-stakes presentations where formatting must remain locked and professional.

This shift ensures that the work done within the ChatGPT ecosystem doesn't stay trapped there. You can take your AI-generated deep dive and turn it into a boardroom-ready document in seconds.

Availability: Who Gets it When?

OpenAI is following a tiered rollout strategy for this update, ensuring that their highest-paying customers get the first look at these premium features:

The mention of the ChatGPT Go tier is particularly interesting. This newer subscription level, designed to bridge the gap between Free and Plus, underscores OpenAI's strategy to bring advanced agentic tools to a broader audience without the $20/month price tag of the Plus plan.

Final Thoughts: The Future of Agentic Work

This update represents a shift in how we interact with LLMs. We are moving away from the "prompt and response" era and into the "collaborative agent" era. With the full-screen viewer and the refined intelligence of the GPT 5.2 model, ChatGPT is proving that it isn't just a toy for writing poems or summarizing emails—it is a legitimate tool for the information age.

By giving users the steering wheel and a clear dashboard to view the road, OpenAI has made Deep Research more than just a search engine; they’ve made it an extension of the modern professional's mind.