Just two months after the launch of the Gemini models, and two days after announcing the expansion of its access to more countries, Google has released Gemini 1.5 this week and made it available to developers and enterprise users for the first time, before launching it publicly for all consumers.
“Our teams continue to expand the boundaries of our latest models with security as the central focus. They are progressing rapidly. In fact, we are ready to introduce the next generation: Gemini 1.5. The 1.5 Pro version offers comparable quality to the 1.0 Ultra version with lower resource consumption“, writes Sundar Pichai, CEO of Google, in the announcement post.
“It also introduces an innovative experimental feature of long-context comprehension“, Pichai continues, mentioning the new feature of Gemini 1.5 Pro. “Gemini 1.5 Pro comes with a standard context window of 128,000 tokens. But, starting today, a limited group of developers and enterprise customers can test it with a context window of up to 1 million tokens through AI Studio and Vertex AI in private preview.”
“Gemini 1.5 offers dramatically improved performance. It represents a radical shift in our approach, based on research and engineering innovations in almost every aspect of the development and infrastructure of our baseline models”, adds Demis Hassabis, CEO of Google Deepmind. “This includes making Gemini 1.5 more efficient for training and serving, with a new Mixture-of-Experts architecture (MoE)”.en.
