Google plans to release Gemini 3 Pro Preview in November

Google’s Gemini 3 Pro preview model has now appeared in VertexAI code, pointing to a launch window that aligns with earlier rumors of a November reveal. The preview is labeled “11-2025,” strongly indicating it will become accessible to select users within this month, while a broader release is likely scheduled for December. This timeline matches prior A-B testing activity observed in AI Studio, where the model has been quietly benchmarked by Google across various scenarios. For developers and enterprise teams already building with VertexAI, the Gemini 3 Pro preview means early access to Google’s next flagship multimodal model, potentially featuring a 1 million token context window, maintaining the scale set by previous Gemini iterations.

The functionality and reach of Gemini 3 Pro are particularly relevant as Google integrates these models into both its API services and broader product suite. The confirmation of a 1M context window would continue to position Gemini models as strong candidates for handling extensive documents and data pipelines, appealing to AI researchers, startups, and enterprises requiring high capacity for context-aware tasks.

The emergence of Gemini 3 Pro also intersects with recent findings about Nano Banana 2, Google’s forthcoming image generation model, which is rumored to be announced soon. There is ongoing speculation about whether Nano Banana 2 will leverage Gemini 3 Pro under the hood, or if the two will be rolled out together as part of a larger multimodal update across Google’s AI stack. The features were uncovered through unreleased code, suggesting that their debut is imminent. For now, Google has not issued any official statements, but these back-to-back discoveries indicate a pivotal moment for the Gemini roadmap, with more clarity expected in the coming weeks.