Bengaluru Startup Releases Open-Source LLM Trained On 22 Indian Languages
The model, released under a permissive license, posts state-of-the-art results on Indic benchmarks while running on a single consumer GPU.

BENGALURU — The launch marks a significant milestone for India's nascent foundation-model ecosystem, which has long depended on Western releases retrofitted for local languages. Key takeaways are covered across 2 sections below, with quotes and context from our newsroom.
BENGALURU — The launch marks a significant milestone for India's nascent foundation-model ecosystem, which has long depended on Western releases retrofitted for local languages.
Early benchmarks show competitive performance in summarisation and reasoning across Hindi, Tamil, Bengali, Marathi and Urdu.



