I’ve been using ChatGPT Deep Research for all sorts of tasks that require a more thorough search of the internet for answers, and I’ve been mostly happy with the results. The AI puts out detailed reports about the topic, and it includes sources to verify its accuracy. For example, I used ChatGPT Deep Research to visit Tokyo and wouldn’t do it any other way. I’ll do the same thing when I travel elsewhere.
My only worry is that I’ll reach the ChatGPT Deep Research limits before the end of each month. I pay for ChatGPT Plus, and my subscription tier only includes 10 Deep Research queries a month. Well, that used to be the case until this week’s big upgrade.
OpenAI decided to bring the Deep Research feature to ChatGPT Free users and raised the limits for premium accounts at the same time. That’s great news, but it comes with a caveat. Not all these Deep Research reports will be the same quality, and that’s something that users will definitely need to keep in mind.
“We’ve noticed many of you love using Deep Research, so we’re expanding usage for Plus, Team, and Pro users by introducing a lightweight version of deep research in order to increase current rate limits,” OpenAI said on X. “We’re also rolling out the lightweight version to Free users.”
OpenAI explained that the lightweight version of Deep Research will be powered by the ChatGPT o4-mini reasoning model, which was unveiled a few weeks ago. The lightweight version is “nearly as intelligent as the Deep Research people already know and love while being significantly cheaper to serve.” o4-mini Deep Research will be “shorter while maintaining the depth and quality you’ve come to expect.”
The company also explained that the original Deep Research will take precedence. All your Deep Research requests will initially use the original, stronger model. Only when you’ve run out of initial Deep Research allotment will ChatGPT revert to o4-mini Deep Research reports.
ChatGPT Free users will get 5 Deep Research reports powered by the o4-mini model. OpenAI also listed the new limits for Deep Research for premium tiers, as follows:
- Free – 5 tasks/month using the lightweight version
- Plus & Team – 10 tasks/month, plus an additional 15 tasks/month using the lightweight version
- Pro – 125 tasks/month, plus an additional 125/month using the lightweight version
- Enterprise – 10 tasks/month
The feature is already available to ChatGPT users. I checked my stats, and I currently have 19 remaining Deep Research jobs by the end of the month. That means I still have 4 regular Deep Research reports left and 15 4o-mini Deep Research reports.
While the upgrade is great, and 4o-mini is the precursor of OpenAI’s next-gen big reasoning AI model, it still feels like OpenAI is nerfing the Deep Research functionality by employing a mini model for some queries, all to say it brought Deep Research to the ChatGPT Free tier.
I don’t think they’ll be that bad, but o4-mini Deep Research reports will require testing, of course. Also, remember that o4-mini can hallucinate information at a higher rate than expected, according to OpenAI’s ChatGPT research. I wonder if that’s going to be an issue for Deep Research.
Put differently, I’ll treat Deep Research almost as if nothing had changed. I’ll consider that I get 10 high-quality reports with my subscription, and those will be dedicated to priority topics.
One way to improve this upgrade is to give the user a choice between full-strength Deep Research reports and lightweight ones. But that’s not possible now, as there’s no toggle to switch between Deep Research models.
