प्रारंभ 2026
This Makar Sankranti, as the sun’s position moves north, we will move to the next chapter of India’s energy revolution.
More on January 14th.
Stay tuned!
From booking your holiday rides to generating that perfect Christmas card, Kruti handles the to-do so you can enjoy the to-be.
This Christmas, let your AI Assistant handle the holiday hustle. 🎄
Merry Christmas! 🎁✨
Not sure whether to use AI Pods or GPU VMs?
Rule of thumb:
Explore with Pods. Scale with VMs.
AI Pods start at Rs 24 per hour, billed in 15-minute intervals.
GPU VMs are built for long running, production workloads.
Explore compute options → https://t.co/4dZmkDqNJn
Full GPUs are overkill for most experiments.
With AI Pods you can:
• Start from ₹24/hr and scale
• Pay in 15-min increments
• Pause when idle, resume anytime
• Use pre-configured AI templates
Fast experiments. Zero infra friction.
Explore AI Pods → https://t.co/4dZmkDqNJn
GPU Instance setup shouldn’t be a chore.
On Krutrim, it’s fast, clean, and ready in minutes.
No quotas. No delays. No nonsense.
Just GPUs that are actually ready when you are.
Create one now: https://t.co/4dZmkDqNJn
Your users are in India but do your GPUs sit abroad? That adds latency, compliance risk, and forex chaos. We run A100s and H100s inside India with stable INR pricing and local data paths.
If you want GPU infra built for India, start here: https://t.co/4dZmkDqNJn
When your GPU workloads are ready but your GPUs aren’t, you lose time and money.
Krutrim solves the obvious bottlenecks:
- A100 from ₹105/hr. H100 from ₹165/hr.
- INR pricing, on-demand & reserved, fast spin-up, consistent performance.
Explore GPU VMs: https://t.co/4dZmkDqfTP
Our paper BhashaKritika is accepted at AAAI-26!
India lacks large, clean pre-training data for its languages. So we've built a 540B-token synthetic corpus across 10 Indic languages, with a pipeline focused on quality, coverage, and low bias.
Paper: https://t.co/2KInHTcUcQ
Most speech tools miss Indian accents and code-mixed sentences.
Shruti gets them right — transcribing English and Hinglish audio accurately for calls, meetings, and media.
Try it now on Krutrim AI Studio: https://t.co/sANBspgSrd
Models can be fluent yet still feel “off.” The way people explain, guide, or give advice varies across languages. Pragyaan captures those cultural and conversational patterns in post-training data across 10 Indian languages.
Catch us at EMNLP today!
Paper: https://t.co/OaGUSEZkS4
We're presenting Pragyaan at EMNLP 2025 this Sunday! A human-in-the-loop pipeline for culturally grounded post-training datasets across 10 Indian languages, blending translation, synthetic generation, and human curation.
Paper: https://t.co/OaGUSEYN2w
Kruti just got smarter! With our all-new Task Agent, you can schedule prompts for any date or time, and let Kruti handle it. You can also set recurring reminders for your daily health and routine needs!
Download now👇
iOS: https://t.co/RMCoU96vDv
Android: https://t.co/kH1ep0iya6
Sorry to hear about your experience.
If you could share your chat via DM (using the share icon at the top, or from the … menu on mobile), we can take a closer look.
At first glance, since this happened after a long conversation, it’s possible that an internal limit was reached or the model switched, causing it to lose context.
@Dev1012194 Thanks for your message.
Aksai Chin is part of India but currently occupied by China — that’s India’s national stance, and the AI includes this in its full response. It reflects the ground reality, not a political opinion.