Ok, so for anyone who's upset about the @SarvamAI models not beating GPT5/other 1T parameter models, my response: are you f kidding me? A year ago you were crying about India not having our own LLMs trained from scratch. The team has launched 14 amazing products in 14 days and it took years of work, each. And they span text, OCR, voice, LLMs, STT, hardware! I mean hardware! Edge models! Celebrate. They are just getting started. Sure, I am biased as their first investor, but we invested because we thought so highly of them. Of course we are biased. And the odds were enormous. But the opportunity is enormous-er :-) and these guys are absolute killers, and the team hunts! I haven't chatted with @pratykumar today but I heard someone say he saw that PPT for today's LLM launch session literally ON THE STAGE for the first time and laid it all out without a prompter. You can only do it when you know your material cold, and your heart is 100% living and breathing this stuff. Just incredible, congrats to everyone at @SarvamAI. Start of an era.