OpenAI is bringing its Deep Research tool to free users with a ‘lightweight’ version – Crypto News – Crypto News
Connect with us
OpenAI is bringing its Deep Research tool to free users with a ‘lightweight’ version OpenAI is bringing its Deep Research tool to free users with a ‘lightweight’ version

Technology

OpenAI is bringing its Deep Research tool to free users with a ‘lightweight’ version – Crypto News

Published

8 hours ago

on

OpenAI is now rolling out a new, lighter version of its agentic deep research tool to free users, while also upgrading the limit for Plus, Team, Enterprise and Edu customers. The ChatGPT maker said its new lightweight deep research tool is powered by the newly released o4-mini reasoning model and is designed to be “cost-effective while preserving high quality”.

OpenAI Pro users will get 250 deep research queries per day, while Team, Plus, Enterprise and Edu users will get 25 queries, and Free users will have access to 5 queries.

Once users reach the limit for the full version of Deep Research, their searches will automatically switch to the lightweight o4 mini-powered version.

Talking about the new lightweight version in a post on X (formerly Twitter), OpenAI wrote, “Responses will typically be shorter while maintaining the depth and quality you’ve come to expect.”

While the Microsoft-backed startup didn’t share many details about the new feature, it did show a graph illustrating that the lightweight deep research model has only slightly lower efficiency than the full-fledged model, while being “significantly cheaper to serve”.

What is Deep Research? Why is it important? 

Deep Research is an AI agent that was first launched by Google, and later gained prominence when OpenAI launched a more expansive version in February. Since then, a number of companies have launched similar offerings under the same or slightly different names. The list of chatbots with deep research/deep search capabilities includes Gemini, Grok, Perplexity and Copilot.

In its earlier blog post, OpenAI had said that Deep Research could synthesise hundreds of hours of online sources to produce a report at the level of a research analyst. Elon Musk made similar claims when he unveiled Grok’s deep research mode in February this year.

While the lofty claims of these companies may not always match reality, it is widely accepted that the deep research models provide better results compared to a Google search or even simply asking a chatbot. However, these agents are still prone to hallucination (making things up), just like any other AI offering.

Related Topics:
Continue Reading

Trending