on november 6th, ufukozen, a technologist, published a post yesterday (5 november) reporting thatGoogleThe next-generation flagship was found on the Vertex AI AI Models Gemini 3 Trace of Pro PreviewModel label is "gemini-3-pro-preview-11-2025It's expected to be released in November 2025。
From model labels and available exposure information, Google estimates that the preview version of the model will be made available to selected business clients in November 2025 and may be available to the public in December of the same year。
As the flagship AI series of Google, Gemini aims to integrate a wide range of information, such as text, images, audio and video, and code, while the appearance of Gemini 3 Pro heralds an AI interaction or will usher in major changes。
The most expected breakthrough of the new model is its legendContext window up to 1 million TokenI don't know. This capacity is much greater than the current mainstream model, which means that Gemini 3 Pro will have the ability to handle the “rule changer” level of the super-long text。
NOTE: CONTEXT WINDOW REFERS TO THE LENGTH OF INFORMATION THAT THE AI MODEL CAN " REMEMBER " AND PROCESS IN A CONVERSATION OR TASK。1 million Tokens, equivalent to a 300-page book (approximately 750,000 English words)This means that the model can read and analyse a thick book at once, without forgetting the beginning。
In particular, it allows for a one-time analysis of the entire book, complex legal documents or research papers, an in-depth understanding of the large code library and the maintenance of logical coherence in a long, multi-step dialogue. This feature will greatly enhance its efficiency and accuracy in tasks such as advanced data analysis, complex information extraction, etc。
In terms of multi-modular capabilities, Gemini 3 Pro will build on previous generations ' models. It is assumed that it will have more sophisticated video analysis of images, more accurate voice recognition and generation capabilities, and a greater cross-modular reasoning capability, which will better connect and understand the deep linkages between different data types。
In addition, the recent exposure of the new Google Image Generation Model Nano Banana 2 could also be published in collaboration with Gemini 3 Pro, which combined or would achieve unprecedented results in such areas as the graphics and the graphics, demonstrating the strategic intent of Google to build a comprehensive AI ecology。