OpenAI upgrades ChatGPT Deep Research with GPT-5.2

OpenAI has given its Deep Research tool inside ChatGPT a meaningful upgrade. The feature is now powered by GPT-5.2, bringing tighter controls over how research is conducted and how findings are presented, all without forcing users to hop between tabs or tools.

Deep Research, introduced last year as ChatGPT’s more agentic, multi-step research mode, was designed for heavier lifts. Think market scans, academic overviews, product comparisons and campaign performance metrics. With ChatGPT-5.2 under the hood, it is less about raw intelligence and more about precision and user control. Here’s what it brings.

Research that feels less black box

The biggest change is how much more say users now have in shaping the research process. You can connect supported apps within ChatGPT, search specific websites, and even restrict queries to selected domains or defined time ranges.

Filters and comparison parameters can be applied upfront, which is useful if you are narrowing down suppliers, tracking coverage over a set period, or comparing datasets side by side.

There is also a more visible sense of progress. Research sessions now show real-time tracking, and you can interrupt an active run to add follow-up instructions or introduce new sources. That makes the process feel less like waiting for a sealed envelope and more like collaborating with a diligent assistant.

ALSO READ: ChatGPT Deep Research now connects with GitHub

The output side has matured too. Users can generate structured reports or tables directly within the chat, include charts and visual summaries, and switch to a fullscreen report mode for easier reading. It is clearly aimed at people who need something presentable, not just paragraphs of text.

Reports with receipts

OpenAI is leaning into transparency here. Deep Research reports now combine analysis with documentation in a single integrated output. Expect key findings, summarised insights, visual comparisons, and trend charts where relevant. There are also methodology notes explaining how results were compiled, alongside commentary sections for context.

ALSO READ: ChatGPT 5.2 lands with big upgrades for professional work

Crucially, source citations are embedded, making it easier to verify claims or trace information back to its origin. Reports stay within the ChatGPT workflow, so you can review, refine, and export without juggling external documents.

The GPT-5.2-powered Deep Research update is rolling out from today. Will you try it out? Let us know in the comments below.