Google has released a “reimagined” version of its Gemini Deep Research Agent, giving developers access to the tech giant’s most advanced autonomous research capabilities.
Powered by Gemini 3 Pro, the updated system transforms Deep Research from a specialized, report-writing assistant into an autonomous research agent designed to perform long-term reasoning and complex analysis.
In short, the renovated system is capable of analyzing and summarizing massive amounts of complex data and providing more detailed and accurate reports than before.
Google said in a blog post The revised system is also designed to reduce hallucinations, a critical requirement for agents working long periods of time.
“(This) agent is optimized for long-running context gathering and synthesis tasks,” Google said. “By scaling multi-stage reinforcement learning for search, the agent autonomously navigates complex information scenarios with high accuracy.”
“Deep Research plans its investigations iteratively – it formulates questions, reads results, identifies knowledge gaps and searches again,” the tech giant said.
This release allows developers to embed Google’s research capabilities directly into their own applications through the company’s new Interaction API. This interface is designed to allow interaction with different agents and give developers more control amid the increase in agentic AI.
According to Google, early users are already applying Gemini Deep Research in areas such as financial services, biotech and market research, as well as drug toxicity safety research, where accuracy and traceability are essential.
Google also said it will integrate its new deep research agent into its services, including Google Search, Google Finance, its Gemini app, and NotebookLM.
Along with the launch, Google also unveiled a new benchmark for complex web searches. The framework, named DeepSearchQA, evaluates agents on “complex, multi-step information seeking tasks.”
Using benchmarks, Google said it found “significant” performance gains when agents were given more time to search and take logical steps. This is something the company said it plans to develop further in future releases.
The same day news came that OpenAI Unveiled GPT-5.2 has been described as “the most capable series yet for professional knowledge work”.
The simultaneous release underscores the growing competition between the two companies as they race to define the next generation of agentic AI systems.
