Gemini Being Trained to Understand Over 100 Indian Languages: Google Deepmind exec

Behind the scenes, the team behind Google’s AI chatbot Gemini, has been diligently working to expand its capabilities, enabling it to understand over a hundred Indian languages, said Manish Gupta, Senior Director at Google DeepMind. As of today, Gemini, already functions in nine Indian languages, showcasing its growing accessibility across the country, he added.

“Our benchmarks show that it is the best model, bar none, when it comes to a whole range of around 29 Indian languages that we have benchmarked. And many of these capabilities are now carrying over to other languages like various Southeast Asian languages like Mahasa, Vietnamese, and so on,” added Gupta.

He also mentioned that Gemini 2.5, Google’s latest generation AI model, is currently regarded as the best reasoning model available, topping various leaderboards and earning acclaim from researchers, academics, developers, and users alike.

He also added that unlike earlier models, it doesn’t just memorize information but truly understands and connects ideas at a deeper level. The model is also natively multimodal—able to process and interpret text, images, audio, and video.

At the event, Shekhar Khosla, Vice President of Marketing at Google India, highlighted key findings from a study conducted by Kantar and Gemini, which surveyed 8,000 individuals across 18 cities in India. The study highlights that 61% of Indians have never used AI, with 75% of users finding it complex.

Despite this, there's a strong desire among Indians to learn AI. The study also revealed that 76% seek AI assistance with everyday tasks like travel planning.

Gupta and Khosla where speaking at Google Gemini Day that was organised by Google. The event showcased some of the latest advancements in Gemini, including new features like Deep Research, which supports file uploads, improved efficiency, and speed.

Additionally, they highlighted Gemini Veo 2, a video generation model developed by Google DeepMind. If you type a short text description (called a "prompt") in Gemini Advanced, the AI can now turn that into a short, high-quality video that’s 8 seconds long.

This comes at a time when there has been an overall increase in the usage of AI. Tech giants are focussing on increasing features of their respective AI chatbots.

Recently, in a court ruling, it was revealed that Gemini had over 350 million users globally up till March 2025. This update was revealed during an anti trust trial against Google.

After OpenAI launched ChatGPT in November 2022 with support from Microsoft, Google responded by launching its own AI chatbot called Bard in March 2023.

In 2024, Google renamed Bard to Gemini and released a more advanced version. Google said in a blog post that Gemini Ultra scored 90% on a major test called MMLU, which checks knowledge and problem-solving across 57 subjects like math, physics, history, law, and medicine.