Loading video...

Video Failed to Load

Go Home

LiteRT is here. The universal framework for on-device AI. 📱💻 ✨ 1.4x faster GPU performance ✨ Unified NPU acceleration ✨ Seamless PyTorch and JAX support ✨ Ready for Gemma and gen AI Build faster, simpler, and everywhere. Read the blog:

194,484 views • 8 months ago •via X (Twitter)

28 Comments

ikan laut's profile picture
ikan laut8 months ago

This is what I need. To build web, applications, and AI. Then maybe in the future also supercomputers, so that quantum no longer relies on laptops or computers, but only in handheld devices such as smartphones, smart glasses and AR/VR/XR/MR devices.

Ankit Yadav's profile picture
Ankit Yadav8 months ago

Can't wait to try it in the development

Senior CRUD Monkey's profile picture
Senior CRUD Monkey8 months ago

Define "universal"

Respo's profile picture
Respo8 months ago

@grok what is this new LiteRThow is this useful is it related to AI

Moon 🎑's profile picture
Moon 🎑8 months ago

Smartphone can now be automated, ready to take tasks into action.

Adi | Resolved Markets's profile picture
Adi | Resolved Markets8 months ago

The real win here is unified NPU acceleration across different hardware. Means devs can finally stop writing separate optimization paths for every chipset vendor.

Amit Kairon's profile picture
Amit Kairon8 months ago

Insane, this is going to shake things up..waiting for devices to get ripped up.

CEO del Socialismo de Mercado🌹🕊️ 市场社会主义CEO's profile picture
CEO del Socialismo de Mercado🌹🕊️ 市场社会主义CEO8 months ago

@ecuarauz

10turtle's profile picture
10turtle8 months ago

On-device AI frameworks like this can be a big win for businesses better performance, lower latency, and more privacy. That combination matters for real-world adoption.

Адам Прокопович's profile picture
Адам Прокопович7 months ago

Hello. I was banned from Google AI Developers Forum. Can someone review my account and lift the unreasonable suspension?

Aayush Pathak's profile picture
Aayush Pathak8 months ago

@grok how is this different from google ai edge

sol's profile picture
sol8 months ago

@CM_CryptoMatrix 变化真的很快

Akshay Bhasme's profile picture
Akshay Bhasme8 months ago

Hi @googledevs @GoogleIndia @GooglePlay — your partner @BillDesk_India is having a serious connectivity issue. I’ve been unable to complete my onboarding for 3+ days. The form won’t proceed at all. Please look into this.

Sriram Kiron's profile picture
Sriram Kiron8 months ago

awesome

OOM's profile picture
OOM8 months ago

I love any advances in on device tech

vivek nakalgaonkar's profile picture
vivek nakalgaonkar8 months ago

@grok is it lighter than gema?

Kanthipriya Mahale's profile picture
Kanthipriya Mahale7 months ago

On-device AI is underrated. I built Aiden — a fully offline executive assistant for iOS using Core ML. No cloud. No external servers. Everything runs locally. Privacy-first AI is the future.

🌊 Sorbus's profile picture
🌊 Sorbus8 months ago

There will be optimization for others Hexagon NPU?

Saïd Aitmbarek's profile picture
Saïd Aitmbarek8 months ago

we need more on-device/light frameworks & models, let's go!

ZETIC's profile picture
ZETIC8 months ago

@grok does LiteRT run ANE on iOS devices?

Sebastian Buzdugan's profile picture
Sebastian Buzdugan8 months ago

unified npu support with jax is super interesting

Delight Okechukwu's profile picture
Delight Okechukwu8 months ago

Faster GPUs and NPU support makes this very interesting for production use.

Chathebots's profile picture
Chathebots8 months ago

@grok comment esseyer ?

Musty's profile picture
Musty8 months ago

Hey I want to do an internship in your organization

Emma's profile picture
Emma8 months ago

Is there a way to turn it off?

DreamCode's profile picture
DreamCode8 months ago

cool, research it.

Peter's profile picture
Peter7 months ago

How to get these tools more popular to the Google fáns! Vids Stitch Flow Dopple Opal NotebookLM Whisk

Mark Okafor's profile picture
Mark Okafor8 months ago

AI tip of my fingers 👍🏿

Related Videos