Google DeepMind

Today, we’re making Gemini 2.0 available to everyone - and expanding the model family. ✨

Say hello to:

🔵  An updated 2.0 Flash, which is optimized for high-volume, high-frequency tasks at scale - enabling developers to start building production applications.

You can now use it via the Gemini API in Google AI Studio, Google Cloud’s #VertexAI platform and the Gemini app.

🔵  2.0 Pro Experimental: our best model yet for coding tasks and complex prompts.

With a 2 million token context window, it’s able to analyze and understand large amounts of information.

🔵 2.0 Flash-Lite, our most cost-efficient AI model yet - now available across Google products.

It has better quality than 1.5 Flash, at similar cost and speed, and comes with a 1 million token context window, multimodal input and text output.

🔵 And 2.0 Flash Thinking Experimental now available in the Gemini app.

Find out more → https://goo.gle/3CMXg5p

We’re putting our most advanced AI for weather forecasting into more people’s hands. ☁️🌐

Scientists can now access Google’s WeatherNext models as well as real-time and historical data to accelerate research, improve how we respond to disasters and much more. → https://goo.gle/40BBEkp

We’re presenting GenCast: our new AI system for predicting 15-day forecasts – covering everyday weather as well as extreme events. ☁️⚡

How could this tech benefit the world? ↓

Published today in Nature Magazine, GenCast is an ensemble model - meaning that it predicts multiple possible scenarios and shows us how likely different conditions are to occur.

It can:
🌀 Generate a forecast in 8 minutes on a single TPU chip. 
🌀 Deliver superior predictions of extreme weather.
🌀 And outperform ENS - the current best operational global forecast.

GenCast is part of Google’s growing suite of AI-based weather models. To help accelerate progress in this area we’ve made the GenCast model code available to the wider community.

Find out more. ↓ https://goo.gle/41k90X4

We’re thrilled that #AlphaFold has been named one of TIME’s best inventions of 2024. 🧬

Our breakthrough model has been used by 2️⃣million scientists in over 190 countries. Find out more about our work ↓

https://lnkd.in/ee-p5_Pw

Congratulations to Demis Hassabis and John Jumper, who have been named Citation Laureates by Clarivate for Academia & Government, for their teams’ breakthrough work on AlphaFold. 🧬

“I’ve dedicated my career to AI because of its potential to advance science and improve billions of lives, and AlphaFold is the first proof point of this promise. AlphaFold has been used by over 2 million researchers to advance critical work, from enzyme design to drug discovery.”
- Demis Hassabis, CEO and co-founder Google DeepMind

“I’m deeply passionate that we’re able to make the work of scientists faster so medicine and science can work better for society.”
- John Jumper, Director, AlphaFold Team Lead, Google DeepMind


Find out more about this recognition → https://lnkd.in/eJQiNJ7T

We developed AlphaChip: an AI approach for designing microchips. Here’s how it’s helping unlock new technologies. ⚡

🔲 It was used in the latest generations of Google’s Tensor Processing Units (TPUs), which are specialized chips for building AI models.

🔲 This includes Trillium, its sixth-generation TPU, which set the new state of the art - delivering nearly five times the peak performance, double the bandwidth, and 67% better energy efficiency than its predecessor.

🔲 Our ultimate vision is using AI to create chips which are faster, cheaper and more sustainable.

🔲 This could help develop new AI-powered technologies in devices such as smartphones, medical devices and agricultural sensors.

Find out more → https://dpmd.ai/4gItIFw