In the intense product iteration competition in the field of artificial intelligence, Google seems to be preparing for its next major move. According to the latest news, Google plans to launch a new Gemini 3.5 Pro model on July 17th. Compared to the previous lightweight version, Gemini 3.5 Flash, the upcoming Pro model is positioned more高端, aiming to handle more complex and in-depth reasoning tasks and intelligent agent workflows.

image.png

One of the core highlights of this update is its support for an ultra-long context window of up to 2 million Tokens. This capacity improvement not only means the model can process massive documents, code repositories, or long videos at once, but also provides a solid underlying support for building long-duration, continuous intelligent agent tasks. In addition, Google has introduced a new "deep thinking" reasoning mode for the model, aiming to enhance the model's ability to perform more in-depth and organized analysis when facing logically rigorous and structurally complex tasks.

image.png

In direct competition with its rivals, Gemini 3.5 Pro has also demonstrated promising potential. According to test feedback shared on social media, in a coding task involving SVG graphic generation, Gemini 3.5 Pro showed performance comparable to that of Claude Fable5High from Anthropic in terms of the delicacy of interface design, accuracy of graphic generation, and the experience of autonomous intelligent agent workflows.