🧙🏼 Gemini 2.0 Pro (2M token context window)

Howdy, wizards.

There was less big news and hype in AI this week than it’s been for a while. IMO, a nice breather so we can actually play and tinker with these new tools, without the allure of new things being constantly launched.

Here’s what’s brewing in AI.

DARIO’S PICKS

1. Google’s releases Gemini 2.0 lineup, including an experimental Pro model

Google took the spotlight this week, releasing its strongest lineup of AI models so far.

Google released an experimental version of Gemini 2.0 Pro:

• Gemini 2.0 Pro is Google’s best model for coding and complex problems to date, and has a massive 2 million token context window. That’s equivalent to about 25 books, or twice the entire Harry Potter series 🧙🏼‍♂️

• Performance wise 2.0 Pro is a strong model, yet is far from blowing existing frontier models out of the water. Some say, a bit underwhelming.

• If you’re a Gemini Advanced user, you already have it in your Gemini app. For everyone else, you can try it for free in Google AI Studio.

Google also updated their Flash series of models. These models are great “workhorse” models designed for high-volume, high-frequency tasks, with a 1M context window:

• Gemini 2.0 Flash is now available to everyone via API, meaning developers can start building apps based on the model. For many devs, it’s a first choice for generic AI use cases, due to the combination of good performance, fast speed, low cost and lots of memory/context.

• Gemini 2.0 Flash-Lite was also released and it’s Google’s most cost-efficient model to date. How cheap? It can generate a one-line caption for 40,000 photos for less than $1.

Right now, you can upload documents and images to all the models, but they’ll only be able to give you a text output. Google says it has more modalities, including image generation and text-to-speech planned for the models in the coming months.

‎ Why it matters‎ ‎ Google's upgraded lineup brings practical advantages on two fronts. The Pro model's 2M context window enables better analysis of lots of info at once; useful for cases where you have lots of docs, full codebases, knowledge bases, etc.

Meanwhile, I can see the new Flash model in particular becoming highly popular for many types of applications due to cost: it has a cheap, simplified pricing structure where there’s no extra charges extra for long context requests.

TOGETHER WITH FELLOW

The #1 AI Meeting Assistant

Typing manual meeting notes drains your energy. Let AI handle the tedious work so you can focus on the important stuff.

Fellow is the AI meeting assistant that:

✔️ Auto-joins your Zoom, Google Meet, and Teams calls to take notes for you.
✔️ Tracks action items and decisions so nothing falls through the cracks.
✔️ Answers questions about meetings and searches through your transcripts, like ChatGPT

Try Fellow today and get unlimited AI meeting notes for 30 days.

Get Started Free

DARIO’S PICKS

2. OpenAI introduced data residency in Europe

Here’s a small piece of impactful news that went largely unnoticed this week. OpenAI just made it possible for ChatGPT Enterprise & Edu, and API customers, to choose having their data processed and stored in Europe. This means European companies can use and build solutions with OpenAI’s tech now without breaking data sovereignty laws, which are some of the strictest worldwide.

For ChatGPT, that means all convos, files you upload, memory, etc. is processed and stored on European servers. For the API, the choice of European data residency has to be selected upon creation of a new projects, ie you can’t change into EU residency once a project is started.

By the way, OpenAI doesn’t train on customer data for its business plans or when using the API — data residency is all about where the data is processed and stored.

‎ Why it matters‎ ‎‎ EU has entered the chat. I mean, they were already in the chat—here’s a list I made of 91 European companies using AI—but this makes it easier for them to stay compliant.

Ofc, this move is profit thinking by OpenAI but as a European, I gotta say I’m glad the AI companies are adapting to us, despite the over-regulated environment that’s currently leaving Europe in the dust when it comes to tech innovation.

PRODUCT REVIEWS

Tool Picks

The very best AI tools we’ve tried recently. We personally test and review AI tools to find the best one for every task. How we test & review tools.

LEARNING RESOURCES

Brain fuel

A round-up of recently published practical AI resources—podcasts, guides, and how-tos—to help you sharpen your skills.

• Podcast with an OpenAI researcher on why soft skills are the future of work

• The Verge’s hands-on test of ChatGPT Deep Research

• Ethan Mollick’s updated guide on which AI to use right now

• Min Choi’s breakdown of some ways people are using the new o3-mini (mostly coding related)

THAT’S ALL FOLKS!

Affiliate disclosure: To cover the cost of my email software and the time I spend writing this newsletter, I sometimes link to products and other newsletters. Please assume these are affiliate links. If you choose to subscribe to a newsletter or buy a product through any of my links then THANK YOU – it will make it possible for me to continue to do this.