Google unveils Gemini 2.5 updates for enhanced AI on Vertex

Google has introduced enhancements to its Gemini 2.5 Flash and Pro AI models, expanding capabilities on the Vertex AI platform for organisations seeking more sophisticated and secure AI-driven applications.

The latest updates to Gemini 2.5 Flash and Pro models focus on three principal areas: providing more transparent reasoning with 'thought summaries', introducing a new Deep Think mode for advanced problem solving, and strengthening protection against indirect prompt injection attacks.

The 'thought summaries' feature is designed to improve clarity and auditability of enterprise AI systems. It systematically organises a model's raw thoughts, including key details and tool usage, into a clear format. The company said this would allow customers to validate complex AI tasks, ensure alignment with business logic, and simplify debugging processes. The aim is to build systems that are more trustworthy and dependable, addressing a key challenge in enterprise-scale AI deployments.

For complex use cases such as mathematics and programming, Gemini 2.5 Pro is introducing an enhanced reasoning mode called Deep Think. This feature enables the model to consider multiple hypotheses simultaneously before producing a response. Utilising new research techniques in parallel thinking, Google intends for this to help in highly complex scenarios. Gemini 2.5 Pro Deep Think will initially be available to trusted testers via Vertex AI.

Security remains a priority with the updated models. Google has increased Gemini's protection rate against indirect prompt injection attacks during tool use, aiming to make it more suitable for enterprise adoption where security compliance is often critical. The company describes Gemini 2.5 as its most secure model family to date.

Gemini 2.5 Flash will become generally available on Vertex AI in early June, with Gemini 2.5 Pro to follow soon after. Google asserts that these updates will have a tangible impact on business operations, from streamlining processes to improving customer engagement.

Enterprise users have reported efficiencies using Gemini 2.5 on Vertex AI. Mike Branch, Vice President Data & Analytics at Geotab, commented on the balance between performance and efficiency: "With respect to Geotab Ace (our data analytics agent for commercial fleets), Gemini 2.5 Flash on Vertex AI strikes an excellent balance. It maintains good consistency in the agent's ability to provide relevant insight to the customer question, while also delivering 25% faster response times on subjects where it has less familiarity. What's more, our early analysis suggests it could operate at potentially 85% lower cost per question compared to the Gemini 1.5 Pro baseline. This efficiency is vital for scaling AI insights affordably to our customers via Ace."

Gemini 2.5 Pro is positioned as the most advanced model for more intricate enterprise requirements. In addition to Deep Think, it introduces features such as configurable Thinking Budgets, supporting up to 32,000 tokens of processing for finer control over resource allocation and more complex tasks.

Yashodha Bhavnani, Vice President of AI Product Management at Box, described Gemini 2.5 Pro's role in addressing unstructured data: "Box is revolutionising how enterprises interact with their vast, and rarely organised, amounts of content. With Box AI Extract Agents, powered by Gemini 2.5 on Vertex AI, users can instantly extract precise insights from complex, unstructured content – whether it's scanned PDFs, handwritten forms, or image-heavy documents. Gemini 2.5 Pro's advanced reasoning makes it the top choice for tackling complex enterprise tasks, delivering 90%+ accuracy on complex extraction use cases and outperforming previous models in both clause interpretation and temporal reasoning, leading to a significant reduction in manual review efforts. This evolution pushes the boundaries of automation, allowing businesses to unlock and act upon their most valuable information with even greater impact and efficiency."

Diverse organisations, including LiveRamp, are looking to Gemini 2.5 to broaden data-driven capabilities across business lines. Roopak Gupta, Vice President Engineering at LiveRamp, said: "With its improved reasoning capabilities and insightful responses, Gemini 2.5 provides tremendous potential for LiveRamp. Its advanced features can enhance our data analysis agents and add support across our product suite, including segmentation, activation, and clean room-powered measurement for advertisers, publishers, and retail media networks. We are committed to assessing the model's impact across a wide array of features and functionalities to ensure our clients and partners can unlock new use cases and enhance existing ones."

Members of the Google Developer Experts community have also begun building new solutions using Gemini 2.5's enhanced context and reasoning features. Recent examples include a persona-based news recommender for supply chain analysts, a disaster preparedness app that delivers personalised guidance from weather data, and a GitHub Action that automates pull request reviews to identify errors and inconsistencies early in the software development process.

These developments highlight the ongoing efforts by Google to expand the enterprise and developer capabilities within Vertex AI as businesses continue adopting artificial intelligence for a range of applications.