Google DeepMind released Gemini 3.5 Pro on July 17, 2026, marking the company’s most capable AI model to date. The model brings a 2-million token context window, Deep Think reasoning on the $250-per-month Ultra tier, and significant improvements in coding and long-horizon reasoning tasks.

What Gemini 3.5 Pro Changes

The new model surpasses its predecessor with enhanced capabilities for long-form document processing. Users can now pass entire books, codebases, or research papers in a single request. The context window expansion allows developers to build applications that work with substantially more information at once.

Google positioned this release as responding to feedback from enterprise customers and researchers who have pushed the boundaries of what’s possible with existing models. The company emphasized reliability improvements alongside the raw capability gains.

Deep Think Reasoning on Ultra Tier

The Deep Think feature, available exclusively to Ultra subscribers, shifts more computational resources toward reasoning before returning an answer. This mode is designed for complex problem-solving, strategic planning, and tasks requiring multiple steps of logical inference. It trades response latency for accuracy on difficult queries.

Deep Think enables users to offload the mental work of problem decomposition. The model handles breaking down a complex task into subtasks, solving each one, and synthesizing the results back into a coherent answer.

Competitive Timing

The Gemini 3.5 Pro launch arrives amid intense competition in the large language model space. Other AI labs have released comparable models in recent months, and Google’s release signals its commitment to maintaining technological leadership. The July 17 release date coincides with the Shanghai World AI Conference, amplifying the announcement’s global reach.

Google has not announced a free tier for Gemini 3.5 Pro, continuing its strategy of restricting cutting-edge models to paying users.