OpenAI has started offering a cheaper version of its ChatGPT Deep Research tool to all users starting today (April 25, 2025). This new “lightweight” tool helps people search the internet and create detailed reports with sources. According to TechRadar and TechCrunch, the tool will be available across all ChatGPT plans – from free users to paying customers. Enterprise and school users will get access next week.
The lightweight version of deep research is powered by a version of OpenAI o4-mini and is nearly as intelligent as the deep research people already know and love, while being significantly cheaper to serve.
Responses will typically be shorter while maintaining the depth and… pic.twitter.com/H2UD5GThVj
— OpenAI (@OpenAI) April 24, 2025
What is Deep Research?
Deep Research is like having a digital assistant that searches the web deeply for you and writes a complete report with information from different websites. When you ask a complex question, the tool looks for answers online, finds good sources, and puts everything together in an organized report.
The original version uses the powerful ChatGPT 4o model, while the new lightweight version uses the smaller 4o-mini model. This makes it cheaper for OpenAI to run while still giving good results.
How to Use It
To use Deep Research, you simply click the “Deep Research” button in ChatGPT and type your question. The AI will then search the internet and create your report. This process can take several minutes to complete.
Usage Limits by Plan
| User Plan | Original Version | Lightweight Version |
|---|---|---|
| Free | 0 | 5 per month |
| Plus/Teams/Education | 10 per month | 15 per month |
| Pro | 125 per month | 125 per month |
Best Uses for Deep Research
- Literature reviews – finding and summarizing academic papers on a specific topic
- Market research – gathering information about industries or business trends
- Complex questions that need multiple sources to answer completely
Accuracy Concerns
While the tool provides citations, users should be careful about accuracy. A study from the Tow Center found that over 60% of AI search engine citations (including ChatGPT’s) contain errors. OpenAI hasn’t shared specific accuracy rates for Deep Research.
This new lightweight option helps OpenAI compete with Google’s Gemini and other AI research tools while making the technology accessible to more people. The company is clearly trying to balance providing powerful research capabilities with managing the computing costs of running these advanced AI systems.
