Google introduces Gemini 1.5 next-gen generative AI model

Google recently launched the Gemini large language model to drive its AI efforts into different properties and even rebranded the Bard chatbot, which is based on the same model, to just Gemini to make it more identifiable. Now, within two months, the company is already ready with the successor. It is called Gemini 1.5 and it brings “dramatically enhanced performance.” While Gemini 1.0 is still reaching Google’s different applications meant for consumers and businesses, Gemini 1.5 will immediately become available across properties.

The first model to become available under Gemini 1.5 is Gemini 1.5 Pro. It is a mid-sized multimodal model optimised for scaling across a wide range of tasks. It is now in private preview on Vertex AI, and available to developers and enterprise users ahead of a full rollout to consumers. “1.5 Pro introduces a new breakthrough experimental feature in long-context understanding — the longest context window of any large-scale foundation model yet,” Google said. The company claims Gemini 1.5 Pro is on par with Gemini 1.0 Ultra, which is still in preview access.

Gemini 1.5 Pro is built on the Mixture of Expertise (MoE) technique, which allows it to run partly when you ask it a question instead of processing the whole model the entire time. As a result, this approach makes Gemini 1.5 Pro act fast and, at the same time, makes it efficient for Google to run.

This model can manage a context containing one million tokens, meaning 1.5 Pro can process vast amounts of information in one go, including “1 hour of video, 11 hours of audio, codebases with over 30,000 lines of code or over 700,000 words”. With 1.5 Pro, enterprises can accurately analyse an entire code library in a single prompt, analyse and compare content across hours of video, enable chatbots to hold long conversations without forgetting details, etc.

Google also said it is bringing more Gemini models to customers with new updates and expanded availability on its Vertex artificial intelligence (AI) platform. Gemini 1.0 Pro, a model for scaling across AI tasks, is now generally available to all Vertex AI customers.

“Starting today, any developer can start building with Gemini Pro in production. 1.0 Pro offers the best balance of quality, performance, and cost for most AI tasks, like content generation, editing, summarisation and classification,” Google said in a blog post.

In addition, the company’s most sophisticated and capable model for complex tasks — Gemini 1.0 Ultra, is now generally available on Vertex AI for customers via ‘allowlist’. According to the company, 1.0 Ultra is intended for complex tasks, with a particularly strong performance in areas such as complex instruction, code, reasoning and multilingualism, and is optimised for high-quality output.

— Written with inputs from IANS