I know I talk about Deep Research from OpenAI a lot, but every once in a while it still does something that absolutely amazes me. Right now, I currently have it running to analyze and look for an appropriate gift that I want to give a friend of mine. It’s still going—been running about 10 minutes or so—and it’s looked at 58 sources already.
What’s kind of funny is that as I sit here working on other stuff, I occasionally look over to see its thinking process and what it’s doing. Right now it’s cracking me up because it’s looking at a bunch of different pictures of these items and trying to assess their quality. On one item it says, “considering skipping due to the image, it’s tough to gauge quality.”
What it’s thinking about right now really impressed me. It says: “wondering if removing sx=300 and sy=300 increases width”—basically looking at how it might modify a URL of an image to make it bigger than it normally would be, removing the constraint based on size. It’s sitting there thinking, “Can I modify the URL and get a bigger copy of this picture?”
I’ve had to do Google searches before to find alternative copies of content behind paywalls, etc., but this is the level Deep Research operates at. When people ask me to compare OpenAI’s Deep Research to other companies’ research tools, this is the level of detail and ingenuity that sets it apart. In my opinion, from what I’ve seen in my comparisons, it gives unbelievably high quality results compared to others.
This isn’t a commercial—I have no financial interest in saying this. It just happens that every once in a while Deep Research does something that makes me think, “Man, I love this tool so much.”
