Chariot

Building speech reasoning models.

[Chariot @Chariot_IN]

Today we’re releasing Chariot-1-ASR Realtime.

It transcribes speech as it’s spoken, and it’s more accurate than every streaming and batch system we tested, across 10 of India’s most spoken languages.

It began as a research experiment. We post-trained Chariot-1, the foundation model for speech we’re building, for speech recognition, to see how a general speech model would do against systems built specifically for it.

[Parth Sarthi @parthsarthi03]

Today we’re sharing an early glimpse of our research, and of where we think multimodal AI is going.

Chariot-1 is the model we’re building to understand speech in general, not to do any one task. To test its abilities, we ran an experiment: we took Chariot-1 and post-trained it for speech recognition.

[Chariot @Chariot_IN]

Chariot-1-ASR Realtime is live on the Chariot platform today.

Try it: https://t.co/w60cGBGPJv

Full benchmarks: https://t.co/Gew4w1JlLr

[Chariot @Chariot_IN]

Today we’re releasing Chariot-1-ASR Realtime.

It transcribes speech as it’s spoken, and it’s more accurate than every streaming and batch system we tested, across 10 of India’s most spoken languages.

It began as a research experiment. We post-trained Chariot-1, the foundation model for speech we’re building, for speech recognition, to see how a general speech model would do against systems built specifically for it.

[Chariot @Chariot_IN]

As Chariot-1 gets better at pronunciation, code-mixing and accents, transcription improves with it, including in places we never targeted. The same foundation will power what we release next.

[Suvrat Bhooshan @TheBhooshan]

We ran an experiment: how well does our speech-reasoning backbone do on plain speech recognition?

We took Chariot-1 and pointed it at ASR. The result is Chariot-1-ASR Realtime, live today.

[debjani ghosh @debjani_ghosh_]

Super stoked to see what @parthsarthi03 and the team at @Chariot_IN are building!!!

This is exactly the kind of differentiated innovation we need to see coming out of India — building for our languages, our context and our scale.

Just super proud of our young innovators... they are so not holding back!

And yes, I clearly need higher heels 😛🤣

[Chariot @Chariot_IN]

Chariot-1 TTS is out today. It speaks Hindi and English, and code-mixes between them naturally. Highest perceptual quality, lowest WER, and under 75ms to first byte. We think it's the best model you can use right now.

Please try it out at https://t.co/TcmjmR3VFq

[Chariot @Chariot_IN]

Chariot text to speech - https://t.co/X6jQedxrxM

[Chariot @Chariot_IN]

You know the half-second where a voice assistant makes you wait, and you start talking again, and now you're both talking. That gap is latency. We measured ours at under 75ms, fast enough that it's gone.

What does that mean for you? Your AI agents, chatbots, and voice assistants stop keeping people waiting and start responding almost instantly. Try Chariot today and hear the difference.

Try it today! Link in the comment.

#texttospeech #developer #aiagents

[Chariot @Chariot_IN]

Chariot text to speech: https://t.co/L7U3j0tNnK

[Chariot @Chariot_IN]

Reading a number out loud is something a six-year-old does without thinking.

It's also the thing most text-to-speech models still can't do reliably. Prices, dates, OTPs - all of it needs a hand-built layer of rules bolted on top, and it still breaks.
Chariot reads them the way the six-year-old does. Try it today! Link in the comment.

[Chariot @Chariot_IN]

Watch the full panel discussion and extended audience Q&A here: https://t.co/ndnXbtLE0k

[Chariot @Chariot_IN]

Our session at the Impact AI Summit was completely housefull and ran 30 minutes over schedule – driven by intense audience Q&A on the hard realities of building sovereign AI! 🇮🇳🚀

The mandate from that room is clear: building true frontier intelligence from India is a massive ecosystem play. We can't just import global models and run them on borrowed infrastructure. We have to own the power, the hardware, and the research and talent pipeline.

Hosted by our Co-Founder Suvrat (@TheBhooshan), the dialogue brought together the pioneers actively building these pillars:

⚡️ Shri Ghanshyam Prasad (Chairperson, @CEA_India) on preparing our national grid for gigawatt-scale, green AI power loads.
💻 Tarun Dua (@tarundua81) (Founder/MD, @e2enetworks) on architecting India's sovereign compute backbone and democratizing GPU access.
🎓 Prof. Jayadeva (GSV Chair, @iitdelhi) on the urgent need to integrate industry and academia to fuel our talent and research engine.
🧠 Parth Sarthi (@parthsarthi03) (Co-Founder, Chariot) on leveraging India's unique domain experts to build the complex RL environments necessary for sovereign frontier models.

The foundation for India's sovereign AI future isn't just being discussed – it's actively being built.
Catch the full recording in the replies! 👇

[Chariot @Chariot_IN]

The path to AGI is not just about algorithms, it’s about aligning our energy grids, compute capabilities, academic foundations and the people who build them.

We are proud to host "India’s Path to an AGI-Enabling Ecosystem" at the Impact AI Summit. We are uniting the critical layers of Sovereign AI:

⚡ Energy: Shri Ghanshyam Prasad (Chairperson, @CEA_India)
☁️ Compute: @tarundua81 (CEO, @e2enetworks )
🎓 Talent: Prof. Jayadeva (@iitdelhi)
🧠 Models: @TheBhooshan & @parthsarthi03 (Our Founders)

🗓 Feb 20, 2026 | 09:30 AM
📍 Sushma Swaraj Bhawan, New Delhi

[Chariot @Chariot_IN]

Come meet the Chariot team at the AI Impact Summit in New Delhi all week! 🇮🇳🚀

We’ve been heads-down building the next generation of foundation models, and we are excited to share what we've been working on.

📍Where to find us (Feb 16-20):
Hall 5, Booth 5.5 (NVIDIA AI Innovation Pavilion)
Hall 2, Pod 2P01

[National e-Governance Division @NeGD_GoI]

Catch this exclusive interview with Mr. Parth Sarthi, Co-Founder of Chariot AI, as he discusses how Chariot AI is paving the way for interactive AI

Watch now: https://t.co/tnFPMsn6iW

#IndiaAIImpactSummit2026 #ResponsibleAI #PeoplePlanetProgress @parthsarthi03

[IndiaAI @OfficialINDIAai]

Catch this exclusive interview with Mr. Parth Sarthi, Co-Founder of Chariot AI, as he discusses how Chariot AI is paving the way for interactive AI through solutions designed for natural, empathetic, and emotionally intelligent communication.

Watch now: https://t.co/mu9zfFHY8U

#IndiaAIImpactSummit2026 #ResponsibleAI #PeoplePlanetProgress

@PMOIndia @AshwiniVaishnaw @jitinprasada @PIB_India @SecretaryMEITY @abhish18 @kavitabha @GoI_MeitY @_DigitalIndia @mygovindia @parthsarthi03

[DD News @DDNewslive]

India-AI Impact Summit 2026 | Chariot, India’s frontier AI lab, unveils voice-first AI models that understand and respond with empathy, aiming to transform customer support, emergency helplines, and storytelling.

Co-founder @parthsarthi03 shares insights on how these speech models reason over content, modulate tone, and engage naturally.

WATCH: https://t.co/jHKHL8xlXn

@Munmun_Bhat @Chariot_IN #AIImpactSummit2026 #VoiceAI #MadeInIndia #AI

[Parth Sarthi @parthsarthi03]

I am incredibly excited to introduce Chariot. (@Chariot_in)

Suvrat (@TheBhooshan) and I are working a research lab based in India to research systems that can truly understand, reason, and interact with the world starting with speech. We are one of the four teams backed by the @OfficialINDIAai Mission and the Government of India.

Today, we had the incredible honor of meeting and interacting with the Honorable Prime Minister of India @NarendraModi ji at his residence to explain what we are working on.

While I was explaining our model to him, he zeroed in on the core problem and asked me if the model could discern the intent behind the words to determine the correct tone. With his example of "Ram Naam Satya Hai" vs "Ram Ram”, he asked if the model knows when something is solemn vs casual? Does it understand the weight of what's being said, not just the words themselves?

It's exactly the kind of problem we're obsessing over at Chariot— speech isn't transcription, it's intent, emotion, cultural context, all encoded in how something is said.

Grateful to the India AI Mission for this opportunity. Lots to build. 🇮🇳