Key things to know about Google DeepMind’s latest LLM, Gemini 1.5

Google has unveiled Gemini 1.5, an upgraded version of its Large Language Models (LLMs), following the successful launch of its next-generation AI models a few months ago. The company has focused on improving the capabilities of Gemini 1.0 through rigorous testing and refining, resulting in a significantly improved performance. Here are five key points to understand about Gemini 1.5.

What is Google Gemini?

Google Gemini was developed by Google DeepMind in collaboration with other teams, including Google Research. It is a general-purpose AI that can help build a variety of AI services that can work in many different fields to automate tasks. Released in three sizes – Nano, Pro and Ultra, the new Gemini 1.5 is the latest advancement in AI technology.

Google CEO Sundar Pichai said of Gemini 1.5: “Our teams continue to push the boundaries of our latest models with security at the core. They’re making rapid progress. In fact, we’re ready to introduce the next generation: Gemini 1.5. It shows dramatic improvements in many dimensions, and the 1.5 Pro achieves comparable performance to the 1.0 Ultra quality, but uses less computation.”

Google Gemini 1.5: 5 things you need to know

1. The new Google Gemini 1.5 is part of the Gemini family, the company announced in a blog post. It represents a significant advance in AI model development and showcases its capabilities to businesses and developers.

2. The company says its latest AI model offers significant improvements in processing capabilities, efficiency and long context understanding across methods. Gemini 1.5 is built on Mixture-of-Experts (MoE) architecture, making it more efficient in training and serving.

3. Google says the first model they’re releasing for early testing is the Gemini 1.5 Pro. It has multimodal understanding, which means it can perform advanced analysis and reasoning tasks between video, audio, code, and text. This includes processing huge chunks of data, such as large codebases or long videos.

4. Gemini 1.5 brings a new feature that can handle up to a million tokens. Google says it’s the longest context window for a large-scale foundation model. Google demonstrated Gemini 1.5’s capability by analyzing the Apollo 11 mission transcript, solving coding problems, and even analyzing silent films.

5. Google says its latest AI model has undergone extensive ethical and security testing to ensure responsible deployment. It focuses on representation disadvantages and content security. For those who want to try Gemini 1.5, Google has released a limited preview of Gemini 1.5 Pro with up to a million characters in the context window. It is available to enterprise customers and developers through Vertex AI and AI Studio.