On Wednesday, Google announced several updates to its generative AI Large Language Models (LLMs), such as Gemini and Gemma, at its second I/O Connect event in Bengaluru. The company revealed that developers can now access the expanded context window with up to 2 million parameters on the latest Gemini 1.5 Pro model. Additionally, Google introduced Gemma 2, a language model with up to 27 billion parameters for developers. New language tools from Google DeepMind India were also unveiled, enhancing software development with AI-powered features while emphasizing responsible AI practices.
As part of the Gemini era, Google aims to bring the power of multimodality to everyone, expanding the types of questions that can be asked with advances in long context windows. Currently, over 1.5 million developers globally use Gemini models across various tools, with the Google AI Studio platform being the fastest way to build with Gemini. India has one of the largest developer bases on Google AI Studio.
“At Google, we’ve been investing in AI for over a decade, driving fundamental advancements that have led us to the exciting Gemini era. AI is a powerful enabler and a major inflection point that will require us to continuously reimagine what’s possible,” said Ambharish Kenghe, vice-president of Google. “We’re committed to empowering Indian innovators to harness AI’s full potential, creating solutions that not only address India’s unique needs but also shape the future of AI globally. The opportunities with multimodal, mobile, and multilingual AI are immense.”
Google is democratizing AI for Indian developers by focusing on three key areas: multimodal, multilingual, and mobile AI. The company is collaborating with MeitY Startup Hub to support 10,000 Indian startups in their AI journey through Google Cloud credits, an AI-first programming curriculum, and the launch of a nationwide GenAI Hackathon and AI Startup Bootcamp. Indian developers now have expanded access to Google’s powerful AI models with the 2 million token context window in Gemini 1.5 Pro and Gemma 2, the next generation of open models.
The Google DeepMind India team shared updates to empower developers building language solutions for India, including the expansion of Project Vaani in collaboration with the Indian Institute of Science. This project provides developers with over 14,000 hours of speech data across 58 languages, collected from 80,000 speakers in 80 districts. The team also introduced IndicGenBench, a benchmark to evaluate the generation capabilities of large language models (LLMs) on Indic languages, and open-sourced CALM (composition of language models), a framework that allows developers to combine specialized language models with Gemma models.
Additionally, Google introduced the MatFormer framework, which enables developers to mix and match AI models within a single framework for optimal performance and resource efficiency. “India is at the forefront of the AI revolution, as you can see from the innovation that Indian companies are pioneering. From consumer experiences to agriculture to social enterprises, AI has the power to address some of the biggest challenges of our time across many sectors and industries,” said Seshu Ajjarapu, senior director at Google DeepMind. “At Google, we remain committed to helping developers and entrepreneurs advance and deploy technology that makes a palpable difference in people’s lives.”
Google is also launching Google Wallet APIs to simplify the integration of loyalty programs, tickets, and gift cards. For developers using the Google Maps Platform, India-specific pricing is being introduced with up to 70% lower costs on most application programming interfaces (APIs). The company is collaborating with the Open Network for Digital Commerce (ONDC), offering developers building for ONDC up to 90% off on select Google Maps Platform APIs.
“AI is transforming the very fundamentals of software development, including how developers think about what’s possible and how they build it,” said Jeanine Banks, vice-president and general manager of Developer X at Google. “At Google, we believe all developers are AI developers. We want to make it as easy as possible for them to build with AI across surfaces, be it web, mobile, or Cloud.”
Additionally, Google will soon launch the Agricultural Landscape Understanding (ALU) Research API, a limited-availability tool designed to make agricultural practices more data-driven and efficient.
, such as Gemini and Gemma, at its second I/O Connect event in Bengaluru...){kind=link}