Google's Gemini 3.5 Pro just changed long docs forever

What happened

Google is launching Gemini 3.5 Pro with a reported 2-million-token context window, making it the largest in any current production frontier model. This capacity equates to approximately 1.5 million words, or about 15 full novels, allowing the AI to simultaneously remember and process vast amounts of text. As of June 23rd, the model is available in limited enterprise preview through Google's Vertex AI platform, with general availability expected between then and June 30th.

Alongside the increased context window, Google is also introducing "Deep Think Mode." This reasoning mode enables the AI to process problems more carefully and logically before generating an answer. Deep Think, however, is gated behind the Gemini Ultra subscription tier, priced at $250 per month, indicating a segmentation of AI capabilities for different user needs.

Why it matters

The expanded context window in Gemini 3.5 Pro represents a qualitative shift in how users can interact with AI. Instead of chunking documents or re-uploading information, users can now provide an AI with entire contracts, meeting transcripts, research papers, or months of email chains, and the AI will retain all of it in a single working session. This fundamentally alters the scope of problems AI can tackle effectively, moving beyond fragmented interactions to comprehensive analysis.

The introduction of Deep Think Mode, coupled with its premium pricing, signals a growing segmentation within the AI market. Similar to different classes of airline seating, AI offerings are evolving into tiered services: a free tier, a mid-tier, and now a premium tier with advanced reasoning capabilities. This trend indicates that companies are not just competing on raw intelligence, but also on capacity, reasoning depth, and user experience, differentiating offerings based on specific enterprise or power user needs.

This development positions Google to address a core frustration with current AI tools: their tendency to "forget" earlier information. By virtually eliminating this memory constraint for most real-world tasks, Google enhances the practical utility of AI for professionals in fields like legal, HR, project management, and research, potentially saving hours of work that would otherwise be spent manually managing information or re-prompting the AI.

What to watch next

• How quickly will the public general availability of Gemini 3.5 Pro occur, and what will the initial user feedback be regarding its stated context window capabilities?
• Will other leading AI models respond by announcing or rolling out comparable context window increases in the near future?
• How will the enterprise adoption of Deep Think Mode progress, and will its $250 monthly price point be perceived as value for money by businesses?
• Will Google's ability to package these technical capabilities into a user-friendly product prove successful, particularly for non-technical users?
• What new, high-value use cases emerge as a direct result of being able to process entire long documents or complex multimodal inputs in a single AI session?

What this means for you

Business leaders and operators should prepare to leverage AI for tasks previously deemed too complex or fragmented due to memory limitations. Identify and keep ready your longest, most data-intensive documents—such as multi-year contracts, comprehensive financial reports, or extensive customer feedback archives. Once Gemini 3.5 Pro becomes generally available, these can be uploaded in their entirety for immediate analysis, summarization, or query without the need for manual chunking or repeated uploads. This shift allows for more holistic strategic planning, risk assessment, and operational insights.

For professionals whose roles involve significant reading, summarizing, or data interpretation—including those in legal, HR, project management, and research—the promise of AI-driven efficiency is becoming more concrete. Anticipate using these tools to process entire binders of information, synthesize large datasets, or even explain complex visual data like charts or spreadsheets in plain English. This capability moves beyond marginal time savings, enabling a fundamental restructuring of workflows and freeing up significant human capital for higher-level strategic thinking.

Key takeaways

• Google's Gemini 3.5 Pro features a 2-million-token context window, the largest in any current production frontier model.
• This large context window allows AI to process approximately 1.5 million words simultaneously, eliminating memory limitations for most tasks.
• Deep Think Mode, a new reasoning capability, is available at the $250/month Gemini Ultra tier, signaling AI market segmentation.
• The expanded context window fundamentally changes AI utility, enabling comprehensive analysis of entire long documents in one session.
• Users should identify long documents they avoid, as new AI capabilities will allow quick summarization and analysis of these materials.

FAQ

What is the primary new feature of Google's Gemini 3.5 Pro?

The primary new feature of Google's Gemini 3.5 Pro is its 2-million-token context window. This large capacity allows the AI to process and remember a significantly greater amount of information during a single conversation or task, effectively eliminating the common problem of the AI forgetting what it was told earlier. This context window can handle roughly 1.5 million words, which is equivalent to about 15 full novels.

How does the 2-million-token context window benefit users?

The 2-million-token context window fundamentally changes how users interact with AI. It allows them to upload and analyze entire long documents such as contracts, meeting transcripts, research papers, or extensive email chains all at once. This means the AI can maintain all relevant information simultaneously, providing a more comprehensive and qualitatively different experience compared to current methods that require chunking documents or re-uploading sections.

What is Deep Think Mode and who is it for?

Deep Think Mode is a new reasoning capability launched with Gemini 3.5 Pro that allows the AI to slow down and think through problems more carefully before providing an answer. This mode is designed for more complex problem-solving and is currently gated behind Google's Gemini Ultra subscription tier, priced at $250 per month. This enterprise-level pricing indicates it is aimed at businesses and power users who require advanced, deliberate reasoning from their AI.

When will Gemini 3.5 Pro be generally available?

As of June 23rd, Gemini 3.5 Pro is in a limited enterprise preview through Google's Vertex AI platform. General availability for the public is anticipated in a launch window between June 23rd and June 30th. Users should watch for official announcements within this timeframe to begin utilizing the model's new capabilities.

How does this release change the AI competitive landscape?

This release changes the AI competitive landscape by shifting the focus beyond just making AI smarter to competing on capacity, reasoning depth, and user experience. Google's context window lead sets a new benchmark for how much information an AI can process in one go. The tiered pricing, including the premium Deep Think Mode, also signals a market segmentation where companies compete to offer specific capabilities tailored to different user segments, from free users to high-tier enterprise clients.