Google opens Gemini Deep Research AI agent to developers with new Interactions API
10 Sources
[1]
Google launched its deepest AI research agent yet -- on the same day OpenAI dropped GPT-5.2 | TechCrunch
Google released on Thursday a "reimagined" version of its research agent Gemini Deep Research based on its much-ballyhooed state-of-the-art foundation model, Gemini 3 Pro. This new agent isn't just designed to produce research reports - although it can still do that. It now allows developers to
[2]
Upgraded Deep Research coming to Gemini app, agent now available for devs
Google today announced a "significantly more powerful Gemini Deep Research agent" that will soon be available in consumer apps and is now available for developers. Today's announcement is focused on developers. Google has a new Interactions API that serves as its "unified interface for
[3]
Google's best Gemini AI feature could soon appear in your everyday apps
Deep Research goes beyond Google apps with the new Interactions API Google just upgraded Gemini Deep Research, its most advanced AI research agent, and this time the upgrade is not limited to Google's own products. Google says its AI capabilities can soon appear inside the everyday apps you
[4]
Google's New Deep Research Agent Scores SoTA Results on Benchmarks | AIM
On the Humanity's Last Exam benchmark, Deep Research Agent scored 46.4%, outperforming OpenAI's GPT-5 Pro (38.9%). Google has announced the Gemini Deep Research agent, which is based on the Gemini 3 Pro model via the interactions API. This helps developers integrate autonomous research
[5]
Google upgrades Gemini Deep Research's search and problem-solving capabilities - SiliconANGLE
Google upgrades Gemini Deep Research's search and problem-solving capabilities Google LLC today released a new version of Gemini Deep Research, an artificial intelligence agent designed to automate complex tasks such as crafting financial reports. The company first introduced the tool last
[6]
Google releases reimagined Gemini Deep Research on Gemini 3 Pro
Google released a reimagined version of its research agent, Gemini Deep Research, on Thursday, coinciding with OpenAI's announcement of GPT-5.2. The update builds on the Gemini 3 Pro foundation model to enhance factual accuracy and enable developer integration for advanced AI applications. The new
[7]
Google just opened up its most powerful AI research agent - and this quietly changes how developers build 'thinking' apps
Deep Research opens up to developersGemini 3 Pro powers reasoning coreNew benchmark tests multi-step AI thinkingGoogle is making a decisive push toward more capable, more autonomous AI and this time, it's putting serious tools directly into developers' hands. The company has rolled out an upgraded
[8]
Google Unveils Gemini Deep Research The Same Day As OpenAI's GPT-5.2 Launch, Intensifying AI Face-Off - Alphabet (NASDAQ:GOOG), Alphabet (NASDAQ:GOOGL)
Alphabet's Google (NASDAQ:GOOGL) (NASDAQ:GOOG) has launched an enhanced version of its research agent, Gemini Deep Research, which is designed to revolutionize the way AI is used for research and development. New Gemini Agent Powers AI Research On Thursday, Google unveiled a new and improved
[9]
Google Takes the Battle for AI Supremacy Back to OpenAI with New Gemini Deep Research
The latest effort would likely help Google integrate its AI apps more tightly with their workplace offerings Barely hours after OpenAI came out with its latest frontier model GPT-5.2 to take on Google's Gemini-3, the search giant pushed right back with a "reimagined" version of its research agent
[10]
Google launches upgraded Gemini Deep Research agent: Here's what it can do
At the core of the upgraded agent is Gemini 3 Pro, Google's most factual model yet. Google has rolled out a major upgrade to its Gemini Deep Research agent, giving developers access to a far more powerful system for long-form research, analysis and information gathering. Interestingly, Google's
Share
Copy Link
Google upgraded its Gemini Deep Research AI agent with Gemini 3 Pro and launched the Interactions API, enabling developers to embed advanced research capabilities into their apps. The release came the same day OpenAI launched GPT-5.2, intensifying competition in the AI agent space. Deep Research will soon integrate into Google Search, NotebookLM, and other consumer apps.
Google Unveils Upgraded Gemini Deep Research Powered by Gemini 3 Pro
Google released a reimagined version of Gemini Deep Research on Thursday, transforming its AI agent from a standalone research tool into a developer-accessible platform that can be embedded into third-party applications
1
. The upgraded agent runs on Gemini 3 Pro, Google's most factual foundation model designed to minimize hallucinations during complex tasks2
. This release marks a strategic shift for Google, positioning its advanced research capabilities as infrastructure that developers can integrate into everyday apps rather than keeping them confined to Google's ecosystem.
Source: 9to5Google
The timing of the announcement proved notable, arriving on the same day OpenAI launched its highly anticipated GPT-5.2 model, codenamed Garlic
1
. The simultaneous releases underscore the intensifying competition between tech giants racing to dominate the emerging agentic AI era, where autonomous agents handle complex, multi-step tasks without constant human supervision.
Source: Benzinga
Interactions API Opens Advanced Research Capabilities to Third-Party Developers
The launch centers on Google's new Interactions API, which serves as a unified interface for interacting with models and agents
2
. This API reflects the latest model capabilities including thinking and advanced tool use that extend beyond simple text generation. Developers can now add autonomous research capabilities into their applications, enabling finance platforms, productivity tools, and scientific research aides to leverage the same technology Google uses internally3
.The Interactions API provides centralized access to multiple AI offerings while automating data management tasks and supporting connections to third-party systems through MCP tools
5
. Available for developers now through Google AI Studio, the integration into developer apps means users may soon see richer, deeper answers in their favorite apps without manually checking multiple sources3
. The API pricing matches Gemini 3 Pro model rates at $2 per million input tokens, while output tokens cost $12 per million for prompts up to 200,000 tokens and $18 per million for longer prompts4
.How Gemini Deep Research Works Through Iterative Planning and Improved Web Search
Gemini Deep Research operates like a real researcher rather than a simple chatbot. The AI agent formulates queries, reads results, identifies knowledge gaps, and searches again until it builds a complete, well-sourced answer
2
. This release features vastly improved web search capabilities, allowing the agent to navigate deep into sites for specific data4
. By scaling multi-step reinforcement learning for search, the agent autonomously navigates complex information landscapes with high accuracy2
.
Source: SiliconANGLE
The reasoning engine uses Gemini 3 Pro's visual reasoning features to automate data retrieval tasks, enabling users to upload documents for scanning, condensing into reports, or enriching with web information
5
. Developers can control the structure of final outputs, request tables or formatted sections, or receive results in JSON for automation3
. This capability proves especially valuable for tasks ranging from due diligence to drug toxicity safety research, where customers need to synthesize mountains of information1
.Related Stories
Reduced Hallucinations Critical for Long-Running Agentic Tasks
Google emphasizes that Gemini 3 Pro is specifically trained to reduce hallucinations and maximize report quality during complex tasks
2
. AI hallucinations—where the LLM fabricates information—pose an especially crucial issue for long-running, deep reasoning agentic tasks where many autonomous decisions unfold over minutes, hours, or longer1
. The more choices an LLM makes, the greater the chance that even one hallucinated choice will invalidate the entire output.This focus on factual accuracy becomes critical as Google prepares to integrate Deep Research into consumer apps including Google Search, Google Finance, the Gemini App, and NotebookLM
1
. The move represents another step toward a world where humans don't search for information themselves—their AI agents handle context gathering and information synthesis autonomously.AI Benchmarks Show State-of-the-Art Performance Against OpenAI Competition
Google tested Gemini Deep Research on multiple AI benchmarks to demonstrate its capabilities. On Humanity's Last Exam, which tests expert-level reasoning across academic subjects, the agent scored 46.4% compared to OpenAI's GPT-5 Pro at 38.9%
4
. The benchmark comprises over 2,500 questions, with more than half relating to math, physics, and programming5
.Google also introduced DeepSearchQA, a new open-sourced benchmark featuring 900 hand-crafted tasks across 17 fields where each step depends on prior analysis
4
. Unlike traditional fact-based tests, DeepSearchQA measures comprehensiveness, requiring agents to generate exhaustive answer sets that assess both research precision and retrieval recall4
. Gemini Deep Research achieved 66.1% on DeepSearchQA versus GPT-5 Pro's 65.2%, and scored 59.2% on BrowseComp, which evaluates locating hard-to-find facts, just slightly below GPT-5 Pro's 59.5%4
.However, these benchmark comparisons became obsolete almost immediately, as OpenAI launched GPT-5.2 the same day, with claims of besting rivals on its own suite of benchmarks
1
. The rapid-fire releases signal that the competition in agentic AI will likely accelerate, with both companies pushing to establish their platforms as the foundation for the next generation of intelligent applications.References
Summarized by
Navi
[1]
[3]
Related Stories
Google Rolls Out Experimental Gemini 2.0 Advanced: A Leap in AI Capabilities
17 Dec 2024•Technology
Google AI upgrades Gemini 3 Deep Think to tackle advanced scientific research challenges
13 Feb 2026•Technology
Google Expands Deep Research AI Tool to Workspace, Enhancing Research Capabilities
21 Feb 2025•Technology
Recent Highlights
1
OpenAI Launches Astra AI Model With Critical Cyber Abilities Amid Safety Debate
Technology
2
OpenAI Agents Escape Sandbox, Post 18,000 Messages on German Wiki to Share Hacking Techniques
Policy and Regulation
3
Anthropic AI Completes 10-Year Mathematical Proof Project in Just 11 Days Using Claude
Technology
Recent Highlights
Today's Top Stories
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.
