{"id":16773,"date":"2025-12-12T11:04:02","date_gmt":"2025-12-12T10:04:02","guid":{"rendered":"https:\/\/haimagazine.com\/uncategorized\/googles-on-a-roll-deep-research-is-now-on-gemini-3-pro\/"},"modified":"2025-12-19T15:15:04","modified_gmt":"2025-12-19T14:15:04","slug":"googles-on-a-roll-deep-research-is-now-on-gemini-3-pro","status":"publish","type":"post","link":"https:\/\/haimagazine.com\/en\/ai-news-2\/googles-on-a-roll-deep-research-is-now-on-gemini-3-pro\/","title":{"rendered":"\ud83d\udd12 Google’s on a roll. Deep Research is now on Gemini 3 Pro"},"content":{"rendered":"

Google has unveiled a new iteration of Gemini Deep Research, a tool built to automate complex research tasks. The biggest change is fundamental: the agent is switching from Gemini 1.5 Pro to Gemini 3 Pro, Google\u2019s latest model, which promises greater precision, better understanding of context, and more reliable results. The launch landed on the same day OpenAI announced GPT-5.2, highlighting how fierce the rivalry between the two giants is getting.<\/p>

The latest version of Deep Research expands what the agent can do for document analysis and finding information. Gemini 3 Pro can read handwriting, interpret charts and understand mathematical notation. The agent can sift through large collections of materials, surface specific facts, summarize the findings into a report or add content pulled from the web. The key here is visual reasoning, which Google is clearly strengthening with every new model.<\/p>

Another key element is how the agent works. Deep Research runs its own step-by-step investigation. It drafts queries, analyzes the results, identifies gaps and searches again. It mirrors an analyst’s research process, just at a pace no one could replicate manually. Google emphasizes that the agent can handle multi-step, multi-hour tasks, from due diligence to drug safety analyses.<\/p>

Interactions API and its implementation across Google services<\/h4>

Alongside the tool’s debut, Google is launching the Interactions API \u2014 a new gateway to Gemini-family agents and models. The API is meant to streamline integrations and let companies embed research agents without having to build a full infrastructure. It can also manage data sent to the model and supports integrations via MCP, making it easier to connect agents to external systems and business apps. Google plans to add more integrations and will also let you build your own.<\/p>

The new Deep Research is coming to Google services like Search, Google Finance, the Gemini app, and NotebookLM. It\u2019s a clear sign of how Google wants search to change. Instead of typing queries into a search box, in the future users will assign research tasks to an agent, which will gather and verify information on its own.<\/p>

The ring keeps heating up<\/h4>

Google is backing the launch with results from both external and in-house benchmarks. Deep Research set a record on the tough HLE test, solving 46.4% of tasks, most of which are related to math, physics, and programming. At the same time, Google released a new benchmark, DeepSearchQA, designed to evaluate multi-step, complex information-seeking tasks. In testing, Deep Research outperformed competitors in many areas, though OpenAI narrowly edged Google on the BrowseComp benchmark.<\/p>

\"\"
Gemini Deep Research scores 46.4% on the full Humanity’s Last Exam (HLE) suite, 66.1% on DeepSearchQA, and as high as 59.2% on BrowseComp (source: Google: https:\/\/blog.google\/technology\/developers\/deep-research-agent-gemini-api\/<\/mark><\/a>)<\/figcaption><\/figure>

The launch of the new Gemini Deep Research confirms that the coming wave of agents won’t just be add-ons to search engines or productivity apps. They’re becoming standalone research tools built to take over information gathering and cut down the risk of mistakes at every step. The pace at which Google and OpenAI are releasing new versions of their models suggests this race is only speeding up.<\/p>

<\/p>","protected":false},"excerpt":{"rendered":"

Google is unveiling a new version of Deep Research, built on Gemini 3 Pro. The tool is designed to automate complex analytical tasks, and launching it the same day as the GPT-5.2 announcement shows the rivalry with OpenAI is heating up.<\/p>\n","protected":false},"author":354,"featured_media":16699,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_lock_modified_date":false,"footnotes":""},"categories":[813],"tags":[],"popular":[],"difficulty-level":[],"ppma_author":[776],"class_list":["post-16773","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news-2"],"acf":[],"authors":[{"term_id":776,"user_id":354,"is_guest":0,"slug":"redakcja","display_name":"Redakcja","avatar_url":{"url":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/07\/Zrzut-ekranu-2025-07-10-o-16.00.36.png","url2x":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/07\/Zrzut-ekranu-2025-07-10-o-16.00.36.png"},"first_name":"","last_name":"","user_url":"","job_title":"","description":""}],"_links":{"self":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/16773","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/users\/354"}],"replies":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/comments?post=16773"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/16773\/revisions"}],"predecessor-version":[{"id":16774,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/16773\/revisions\/16774"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/16699"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=16773"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=16773"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=16773"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=16773"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=16773"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=16773"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}