Google's Gemini 3.0 Pro Leaked: What to Expect
References to Google's next-generation AI, Gemini 3.0 Pro, have been spotted by developers in the code for Google Vertex AI. The naming suggests an initial preview release is planned, potentially before the end of November. This approach mirrors the rollout of Gemini 2.5 Pro, which moved from a preview to General Availability status between March and June.
Key Features and Enhancements
The leak reveals several exciting upgrades that bring Gemini in line with competitors like GPT-5 and Grok-4:
- Dual Context Windows: Two versions are planned—a standard 200K token window and a larger 1M token version, aligning with the options available for Gemini 2.5.
- Advanced Tool Integration: Logan Kilpatrick of Google AI Studio confirmed that the model will be able to leverage tools like Python for data analysis and perform sequential web searches to gather and refine information during its reasoning process.
Impressive Coding Prowess
Early testers on platforms like Google AI Studio and LMArena have been impressed by its advanced programming abilities, noting it can generate small arcade games or replicate the user interfaces of Windows and macOS with just a couple of prompts.