What you need to know
- Gemini 3 Pro marks the debut of the Gemini 3 series, enabling Google to reclaim its position at the forefront of AI benchmarks.
- The rollout of Gemini 3 begins today, accessible via the Gemini app for all users and through AI Mode in Google Search for paid subscribers.
- These models are celebrated as Google’s most advanced yet, showcasing unparalleled multimodal understanding and agentic coding capabilities.
In a significant advancement for artificial intelligence, Google has unveiled Gemini 3, the latest iteration of its AI model suite. The standout feature, Gemini 3 Pro, is now available in preview across various platforms, including Google Search, the Gemini app, and Google’s suite of developer tools. Google describes Gemini 3 Pro as a pivotal step towards artificial general intelligence (AGI), claiming it to be the leading model globally for multimodal understanding.
This launch is particularly noteworthy as it represents the first instance of Google introducing a new AI model in Search on the very first day. Users engaging with AI Mode in Search can now benefit from enhanced reasoning and agentic coding abilities, providing assistance with more complex inquiries while generating responses that blend text with interactive visuals. Additionally, the Gemini app will feature a Gemini Agent capable of autonomously managing multi-step tasks.
Following the release of Gemini 2.5 models in March 2025, which initially performed admirably on key AI benchmarks, the landscape shifted as competitors like xAI and Anthropic gained traction. However, Gemini 3 Pro has swiftly reversed this trend, securing the top positions on both the LMArena and WebDev Arena leaderboards immediately after its launch. In comparison to its predecessor, Gemini 2.5 Pro, the new model demonstrates substantial improvements across all major AI benchmarks.
Specifically, Gemini 3 Pro has achieved a remarkable score of 1,501 points on the LMArena leaderboard, surpassing Grok 4.1 Thinking, which scored 1,483 points. For context, Gemini 2.5 Pro previously held third place with a score of 1,452 points. In other evaluations, such as Humanity’s Last Exam, Gemini 3 Pro attained a score of 37.5% without tool usage, a result indicative of PhD-level reasoning, according to Google.
For everyday users, Gemini 3 is designed to introduce a new level of depth and nuance in interactions. Unlike many AI models that tend to cater to user expectations, Gemini 3 Pro aims to deliver responses that are “smart, concise, and direct,” challenging the perception that AI cannot provide users with the information they truly need.
For tasks related to website creation and generative user-interface design, Google asserts that Gemini 3 is its most proficient model to date. This is evidenced by an impressive 1,487 ELO score on the WebDev Arena leaderboard, along with superior performance on Terminal-Bench 2.0 and SWE-bench Verified benchmarks. The capabilities of Gemini 3 Pro extend to building playable 3D video games from scratch, creating websites, and fulfilling various vibe-coding requirements.
To further enhance performance, Google has introduced Gemini 3 Deep Think, which improves results on Humanity’s Last Exam to 41% without tool use. This model is currently undergoing additional safety evaluations and will be available to Google AI Ultra subscribers in the upcoming weeks.
Gemini 3 is now live in Google AI Studio, Vertex AI, Gemini CLI, and Antigravity
Gemini 3 is not only being rolled out to Google’s consumer products but is also becoming available for developers. It can now be accessed in Google AI Studio, Vertex AI, and Gemini CLI, as well as on third-party platforms such as Cursor, GitHub, JetBrains, Manus, and Replit. Most notably, Gemini 3 serves as the foundation for Google Antigravity, a new agentic development platform.
Google Antigravity functions as an AI-powered integrated development environment (IDE). While the concept of an AI IDE is not entirely new, Antigravity distinguishes itself through its advanced AI agents. The new features of Gemini 3, including enhanced reasoning and improved agentic coding capabilities, empower these AI agents with unprecedented permissions.
These AI agents possess direct access to the editor, terminal, and browser within Google Antigravity, allowing them to act as coding partners capable of planning, executing, and validating software tasks independently. The platform utilizes a combination of Gemini 3 Pro, Gemini 2.5 Computer Use, and Nano Banana (Gemini 2.5 Image) to facilitate end-to-end agentic workflows.
Try Gemini 3 now in the Gemini app and Google Search
Gemini 3 Pro is now accessible within Google’s consumer products, including Search and the Gemini app. Users subscribed to Google AI Pro or Ultra can experience it by selecting the “Thinking” model in AI Mode. While it will soon be available to all AI Mode users in the U.S., it currently remains a paid exclusive.
In the coming weeks, subscribers to AI Pro and Ultra will benefit from a new automatic model selector in Search, which will direct more complex queries to the flagship Gemini 3 models while assigning simpler tasks to faster, more efficient models.
The enhanced understanding capabilities of Gemini 3 enable AI Mode to utilize a query fan-out method, allowing for multiple searches simultaneously and revealing new content aligned with user intent. By accurately interpreting the purpose behind a user’s request, Gemini 3 and AI Mode can present relevant information that older models may have overlooked.
AI Mode also harnesses generative UI to craft interactive and engaging learning tools in response to searches, such as images, tables, and grids. It can even take it a step further by coding a custom simulation or creating a tool in real-time, depending on the user’s prompt.
Gemini 3 Pro is now available in Google Search for subscribers and in the Gemini app for all users, marking a significant evolution in the capabilities of AI technology.