×
Google Cloud Next – Gemini 2.5 Pro EVERYWHERE
Join our daily newsletter for breaking news, product launches and deals, research breakdowns, and other industry-leading AI coverage
Join Now

# Google Cloud Next: Gemini 2.5 Pro Everywhere – Key Announcements

Google Cloud Next just wrapped up with an impressive array of AI announcements showcasing Google’s accelerating innovation in the AI space. From powerful new models to groundbreaking agent capabilities, here’s a comprehensive look at what was unveiled.

## Hardware: Ironwood TPU

– **TPU Generation 7 (Ironwood)** coming later this year
– **3,600x performance increase** compared to the first publicly available TPU
– **29x more energy efficient** than previous generations
– Designed to power the next frontier of AI models

## Gemini 2.5 Models

### Gemini 2.5 Pro
– Currently ranked as the “best model in the world” according to Chatbot Arena
– Achieved the highest score ever on “Humanity’s Last Exam” benchmark
– Demonstrated impressive reasoning with a Rubik’s cube simulation that was created in one shot (though the presenter didn’t highlight this zero-shot capability)

### Gemini 2.5 Flash
– A new low-latency, cost-efficient model with built-in reasoning
– Allows users to control the amount of reasoning to balance performance with budget
– Coming soon to AI Studio, Vertex AI, and the Gemini app

## Agent Ecosystem

### Agent Development Kit
– New **open-source framework** for building sophisticated multi-agent systems
– Enables Gemini-powered agents to use tools and complete complex tasks
– Supports the **Model Context Protocol (MCP)** for unified data access

### Agent-to-Agent Protocol
– Allows agents to communicate regardless of their underlying model or framework
– Supported by leading partners across the multi-agent ecosystem
– Compatible with frameworks like Langraph and Crew AI

### Google Agent Space
– Demonstrated interoperability between different platforms (Box and Google Cloud)
– Agents can collaborate across platforms to solve complex tasks

## Media Generation Models

### Imagine 3
– Improved text-to-image model with better detail, lighting, and fewer artifacts
– Delivers accurate prompt adherence for precise creative vision

### V2 (Video Generation)
– Industry-leading video generation model producing up to several minutes

Recent Videos