Thursday, April 17, 2025

Start of archive·May 16

Top story

The Agent RaceGoogle DeepMind Blog

Introducing Gemini 2.5 Flash

Gemini 2.5 Flash introduces a new capability tier: reasoning-on-demand. This hybrid approach lets developers optimize for latency vs. accuracy per query, fundamentally changing how AI applications are architected.

Gemini 2.5 Flash is Google's first fully hybrid reasoning model

Read full story