Henrik Hammer

I work in Google Cloud. I follow #CloudComputing #SocialMediaAnalytics #GoogleCloud #dktech Off work I enjoy #Windsurfing #Skiing

[Google Cloud Tech @GoogleCloudTech]

Our 5-Day AI Agents Intensive on Kaggle wrapped with 1.5M+ participants! We moved beyond chatbots to build agents that reason, plan, and act.

If you missed the live sessions, we’ve transformed the entire curriculum into a permanent Kaggle Learn Guide → https://t.co/3EZYPxQvvP

[Logan Kilpatrick @OfficialLoganK]

Gemini 3 Flash on the @ArtificialAnlys intelligence benchmark, the most cost per intelligence efficient model in the world!!!

[Jeff Dean @JeffDean]

Gemini 3 Deep Think is now available for Ultra users, making available our IMO & ICPC Gold Medal-winning technology.

Deep Think shows improved generalization on difficult benchmarks like ARC-AGI-2, and outperforms Gemini 3 Pro on HLE & GPQA Diamond.

We hope this serves as a useful step toward ever-more-capable scientific reasoning agents.
https://t.co/8QaUosoa98

[Logan Kilpatrick @OfficialLoganK]

Nano Banana Pro demos, showcases, and vibes, live from Mountain view 🍌

https://t.co/wZdEUxTI1z

[Jeff Dean @JeffDean]

Our latest Gemini image generation and editing model is quite good (see the examples in the thread below). It lets you indulge in a bit of creative fun, helps you make new business cards, etc!

Try it out at https://t.co/xR9jN4XeNg

@GeminiApp

[Logan Kilpatrick @OfficialLoganK]

Introducing Gemini 2.5 Flash Image (aka nano-banana), our SOTA image generation and editing model 🍌

As you might have already seen, this model excels at character consistency, creative edits, and has Gemini's world knowledge!

[Jeff Dean @JeffDean]

Starting today, we're offering Deep Think in the Gemini app for Google AI Ultra subscribers, and we're giving a collection of mathematicians access to the full version of the Gemini 2.5 Deep Think model that achieved gold medal🏅 level performance in the recent IMO competition. This model excels at complex math, reasoning and coding.

See more at: https://t.co/g5qw0NTFLL

[1a3orn @1a3orn]

So I ran my bench of perfect-information-games against the new Gemini 2.5 Experimental and -- it's quite good.

It displaces DeepSeek V3, by a lot. But I also found that reading through the transcripts of its thoughts, they seemed qualitatively different.

[swyx @swyx]

With gemini 2.5 pro pricing and results, Google has fixed the biggest unknown/weakest link in their lineup

and we can now confirm that @GoogleDeepMind completely owns the pareto frontier down to 1220 elo

(at least until o4-mini and gpt5?)

[Steven Johnson @stevenbjohnson]

As a longtime "tools for thought" user, I'd seen a lot of mind maps, but I'd never been a big believer.

But in one inspired weekend of prototyping, @HcwXd completely changed my mind.

Mind Maps are now available to all @NotebookLM users. Here's why they're different...

The problem I'd always had with mind maps really came down to two things: they were complicated to build manually, and once you'd assembled them, they weren't all that useful. Like, it was nice to see a diagram of the core concepts in whatever material you were mapping, but once you had that visualization, there wasn't really a next step that helped you explore any of those concepts.

But about four months ago, shortly after he joined the Notebook team, @HcwXd built a mind maps prototype in his spare time. It eliminated the laborious process of manually creating links or tagging with concepts, and just generated an interactive map of all your sources automatically. That was amazing enough.

But the thing that really blew me away was this: every node in the map launched a full NotebookLM query with one click. So you can get an amazing visualization of the core ideas or themes in your sources, but then instantly drill down to get a detailed overview of a specific topic, with citations back to the original source material.

It's a genuinely new way to navigate through a large body of information.

The best way to try it out is to create a new notebook, and add the last 20 or so documents you've written or read recently (PDFs/Docs/Slides/etc.) Then generate a mind map -- you'll have an amazing visualization of where your head has been for the past few weeks.

[Matthew Berman @MatthewBerman]

Gemini 2.5 Pro is insane at coding.

It's far better than anything else I've tested.

Below are ONE SHOT demos 🧵

[Jeff Dean @JeffDean]

🥁Introducing Gemini 2.5, our most intelligent model with impressive capabilities in advanced reasoning and coding.

Now integrating thinking capabilities, 2.5 Pro Experimental is our most performant Gemini model yet. It’s #1 on @lmarena_ai leaderboard. 🥇

[Logan Kilpatrick @OfficialLoganK]

Big updates to the @GeminiApp today:

- Deep Research is now available to all users for free
- Deep Research is now powered by 2.0 Flash Thinking (our reasoning model)
- New Gemini personalization using your search history
- Gems are available to all users

And more in 🧵

[Sundar Pichai @sundarpichai]

2/ We’re introducing a new model Flash-Lite, which is extremely efficient and capable, in public preview.

Plus an experimental version of Gemini 2.0 Pro, our best model for coding performance and complex prompts, is now available. Gemini Advanced users can try it out in the @geminiapp now. Model availability at a glance:

[Sundar Pichai @sundarpichai]

1/ New Gemini 2.0 updates, here we go!

Gemini 2.0 Flash is now GA, so devs can now build production applications. Find it in AI Studio or Vertex AI.

[Poonam Soni @CodeByPoonam]

Google just launched Gemini 2.0 Flash and it's INSANE

This will change the way we use technology forever.

Here’re 5 Powerful features you don’t want to miss:

[Arena.ai @arena]

New Gemini-2.0-Flash-Thinking improved over the previous 1219 thinking checkpoint.

[Urs Hölzle @uhoelzle]

If you have Windows laptops/PCs that have been hacked, or will be hacked soon 🤔: consider doing what Nordic Choice Hotels, a chain with more than 200 hotels, did within 24 hours: switch to ChromeOS. https://t.co/qdWWD9NT8a

[Henrik Hammer @HammersTwit]

Announcing preview of BigQuery’s native support for semi-structured data @googlecloud https://t.co/9YlpSYzdFi

[Urs Hölzle @uhoelzle]

Quartz: Google made clean energy cool, and big batteries are next.

“Any organization could do this,” said Finn-Foley. The [...] hurdles are significant but surmountable. “The advantage Google has [...] is the cultural will to do so.”

https://t.co/6Y21Rq4ORL

via @GoogleNews