The lightweight version of Deep Research for users of the free app uses OpenAI’s o4-mini model which will deliver responses that the AI company says are “nearly as intelligent” as the full version of Deep Research. From OpenAI’s perspective, the lightweight version will be more cost-effective to run. While the company does say that the lightweight results are shorter, they will have the same depth and quality as responses from the full version.
Deep Research analyzes, synthesizes, and combines data from hundreds of online sources and can be used to deliver a customized recommendation for a new car to buy, the purchase of a new major appliance, or a new big-screen television. The feature can also be used to create market analysis, find a place to live, or help select a college to attend.
Free users of the ChatGPT app can use the lightweight version to use Deep Research for no more than five tasks a month. I have a feeling that this will be raised to more uses in due time. Other plans have the following restrictions:
- Plus & Team users: 10 tasks/month (full version) + 15 tasks/month (lightweight version).
- Pro users: 125 tasks/month (full version) + 125 tasks/month (lightweight version).
- Enterprise users: 10 tasks/month (full version).
Users that reach their full version limit will be switched to the lightweight version automatically so as not to face any interruptions. EDU and Enterprise users will be able to begin using the DeepResearch tool next week.
Google’s GeminiAI feature also has a similar version of DeepResearch available for paying members. It also has the goal of coming up with reports and analyses in minutes that would normally take humans hours to complete.
Source link