2 months at Google DeepMind - Recap

A quick summary of all exciting developments and product updates

Hi there👋

It’s been a while since you heard from me. I joined Google DeepMind two months ago, and the weeks have been extremely busy and packed with exciting developments.

Below is a quick summary of everything that happened. As part of the Developer Relations Engineering team for Gemini and Gemma, I've had a front-row seat to most of those launches:

Model updates:

• Gemini 2.5 Pro

• Gemini 2.0 Flash native image generation

• Gemini Robotics

• Gemini Embedding

• Imagen 3

• Gemini 2.0 Flash

• Gemini 2.0 Flash-Lite

• Gemini 2.0 Pro

• Gemma 3

• ShieldGemma 2

• PaliGemma 2 Mix

Product improvements:

• YouTube link support

• Conversation branching in Google AI Studio

• AI co-scientist

• Audio overview & Canvas mode (Gemini app)

• AI Mode powered by Gemini (Google Search)

• Code execution as a tool

Dev Ex:

• Gen AI Python SDK v1.0.0

• Gen AI JS/TS SDK (preview launch)

• Improved OpenAI SDK compatibility layer

• Major docs updates (still ongoing)

• New cookbooks

Hands-on tutorials from me (not official Google tutorials):

• Get started with Gemini's multimodal capabilities

• Image creation and editing with Gemini 2.0

• Create workout summaries with Gemma 3 1B

The pace here is incredible, and moving forward, I plan to share more updates and insights into the exciting work I see at Google DeepMind (but note that I can't talk about unreleased products).

I know that my newsletter hasn't been very consistent, and with this new content shift, I completely understand if you decide to unsubscribe. Thank you all for your past engagement and support! And for those of you who will be staying, my goal is to make this new focus on AI research, product updates, and tutorials genuinely interesting and valuable.

Best,

Patrick