Google DeepMind’s Deep Research Max runs 160 queries per task autonomously

Google DeepMind has unveiled Deep Research Max, a new autonomous AI research agent designed to handle complex, multi-step web research tasks. Now available in public preview through the Gemini API, the agent is powered by Gemini 3.1 Pro, enabling it to execute approximately 160 search queries per task. This capability allows it to explore diverse sources, cross-reference information, and synthesize findings without human intervention. In benchmark tests, Deep Research Max achieved a 93.3% score on DeepSearchQA, a metric that evaluates the quality of web research outputs.

The agent represents a significant leap in automated research, targeting professionals who need to gather and analyze large volumes of online data quickly. By autonomously running multiple search queries and iterating on results, it can produce comprehensive reports on topics ranging from market analysis to scientific literature. Google DeepMind positions this as a tool for knowledge workers, researchers, and analysts to accelerate decision-making. The public preview signals Google’s confidence in the technology and invites developers to integrate it into custom workflows via the Gemini API, with pricing aligned to token usage.