Google launched its deepest AI research agent yet — on the same day OpenAI dropped GPT-5.2

Google launched its deepest AI research agent yet — on the same day OpenAI dropped GPT-5.2

Google launched on Thursday a “reimagined” version of its research agent Gemini Deep Research based on its much-ballyhooed state-of-the-art basis mannequin, Gemini 3 Pro.  

This new agent isn’t just designed to supply research experiences — although it may still do that. It now permits builders to embed Google’s SATA-model research capabilities into their very own apps. That functionality is made potential by way of Google’s new Interactions API, which is designed to offer devs more management in the coming agentic AI period. 

The new Gemini Deep Research tool is an agent outfitted to synthesize mountains of knowledge and deal with a large context dump in the immediate. Google says it’s utilized by clients for duties starting from due diligence to drug toxicity security research. 

Google also says it would soon be integrating this new deep research agent into providers, including Google Search, Google Finance, its Gemini App, and its widespread NotebookLM. This is another step towards getting ready for a world the place people don’t Google something anymore — their AI brokers do. 

The tech large says that Deep Research advantages from Gemini 3 Pro’s standing as its “most factual” mannequin that is skilled to attenuate hallucinations during advanced duties.

AI hallucinations — the place the LLM just makes stuff up — are an particularly essential issue for long-running, deep reasoning agentic duties, during which many autonomous choices are made over minutes, hours, or longer. The more decisions an LLM has to make, the higher the likelihood that even one hallucinated alternative will invalidate the total output. 

To show its progress claims, Google has also created yet another benchmark (as if the AI world wants another one). The new benchmark is unimaginatively named DeepSearchQA and is meant to check brokers on advanced, multi-step information-seeking duties. Google has open sourced this benchmark.  

The Gossip Blogger event

San Francisco
|
October 13-15, 2026

It also examined Deep Research on Humanity’s Last Exam, a a lot more curiously named, impartial benchmark of basic data full of impossibly area of interest duties; and BrowserComp, a benchmark for browser-based agentic duties.

As you may count on, Google’s new agent bested the competitors on its personal benchmark, and Humanity’s. However, OpenAI’s ChatGPT 5 Pro was a surprisingly shut second all the approach round and barely bested Google on BrowserComp. 

But those benchmark comparisons have been out of date nearly the second Google printed them. Because on the same day, OpenAI launched its extremely anticipated GPT 5.2 — codenamed Garlic. OpenAI says its latest mannequin bests its rivals — particularly Google — on a collection of the typical benchmarks, including OpenAI’s homegrown one. 

Perhaps one in all the most attention-grabbing components of this announcement was the timing. Knowing that the world was awaiting the release of Garlic, Google dropped some AI news of its personal.

Stay informed with the latest headlines that matter. At TheGossipBlogger.com, we ship well timed and credible coverage on breaking news, global occasions, politics, society, and every little thing in between.

Whether it’s unfolding developments, coverage modifications, or highly effective human-interest tales, our newsroom curates impactful content to maintain you up to date in real time.

From local points to worldwide affairs, we break down advanced tales with readability, context, and a spotlight on what’s related to you.

Bookmark News and test in often — because staying informed is the first step towards staying ahead.