Google announced Gemini 3 Pro, its most intelligent AI model, a few weeks before the first anniversary of Gemini 2. The model delivers state-of-the-art reasoning and class-leading coding performance. It becomes available immediately in the Gemini app, Google Search’s AI Mode, and developer platforms.
Now this is the best for how long…
Gemini 3 Pro sets a new benchmark record on Humanity’s Last Exam, a test recognized as one of the most challenging evaluations for AI systems. The model achieves 37.5 percent accuracy, surpassing the prior leader, Grok 4, by 12.1 percentage points. This result comes without reliance on external tools such as web search, demonstrating the model’s standalone capabilities in handling complex, diverse questions across subjects.
On the LMArena leaderboard, Gemini 3 Pro claims the top position with 1,501 points. This score reflects its performance across multiple evaluation categories, positioning it ahead of competing models in real-world user preference metrics compiled by the platform.
Immediate availability in the Gemini app
Within the Gemini app, Gemini 3 Pro generates responses that are more concise and exhibit improved formatting. Users access the model by selecting “Thinking” from the model picker, making it available to all users. AI Plus, Pro, and Ultra subscribers benefit from higher rate limits, allowing extended usage before encountering restrictions.
Introducing Gemini Agent for proactive workflows
Gemini 3 Pro enables Gemini Agent, a feature that extends capabilities from Project Mariner, Google’s web-surfing Chrome AI introduced at the end of the previous year. Gemini Agent performs specific tasks on behalf of users. For instance, when managing an email inbox, the agent executes the work directly rather than providing only general advice, such as organizing messages or drafting replies based on user instructions.
To utilize Gemini Agent fully, users grant it access to their Google apps, integrating it seamlessly with services like Gmail and Calendar. This permission allows the agent to interact with personal data and complete actions within those environments.
Enhancements to Google Search’s AI Mode
In Google Search’s AI Mode, Gemini 3 Pro launches first for AI Pro and Ultra subscribers through the “Thinking” dropdown selection. The model extends to AI Overviews, targeting the most difficult queries posed to the search engine. A new routing algorithm for AI Mode and AI Overviews deploys in coming weeks, directing complex questions to Gemini 3 Pro automatically based on query analysis.
Gemini 3 Pro enhances AI Mode’s content discovery by refining the fan-out search technique. This method expands queries into multiple parallel searches. The model’s advanced intelligence conducts additional searches and identifies relevant, credible content that earlier models overlooked, improving result comprehensiveness.
The model’s multi-modal understanding supports dynamic interfaces in AI Mode responses. For mortgage loan research, it generates an interactive loan calculator embedded in the output, enabling users to input variables like interest rates, loan amounts, and terms to compute payments and amortization schedules directly within the search interface.
Developers and enterprise customers access Gemini 3 Pro via the Gemini API, AI Studio, and Vertex AI. These platforms provide APIs for integration, prompt testing in AI Studio, and scalable deployment through Vertex AI for production workloads.
Google released Antigravity, a new agentic coding application. Antigravity programs autonomously, generates its own tasks to advance projects, and delivers progress reports to users. It handles full development cycles, from initial code generation to iterative improvements and status updates.
Introducing Gemini 3 Deep Think
Alongside Gemini 3 Pro, Google introduced Gemini 3 Deep Think, an enhanced reasoning mode. Initial availability goes to safety testers, followed by rollout to AI Ultra subscribers. This mode amplifies reasoning depth for particularly demanding tasks.
Gemini 3 Pro’s deployment spans consumer and professional applications. In the Gemini app, its concise formatting streamlines information presentation, reducing verbosity while maintaining completeness. The integration with Gemini Agent shifts from advisory responses to proactive execution, handling real-world workflows like email triage through direct API calls to Google services.
AI Mode’s fan-out augmentation with Gemini 3 Pro processes queries with greater precision, executing broader search variants to surface authoritative sources. The dynamic interfaces leverage the model’s ability to interpret text, images, and data jointly, producing tools like financial calculators that respond to user inputs in real time.
Antigravity’s autonomous task creation allows it to break down coding projects into subtasks, such as designing algorithms, implementing functions, and debugging, while reporting milestones like completed modules or test coverage percentages. Gemini 3 Deep Think extends reasoning chains, applying iterative thinking to refine solutions before final output.
