Tagged with

4 articles found

Google Casually Dropped the Most Capable Open Weights on the Planet, And Fixed Their License Problem

Gemma 4 arrives with sparse MoE architecture, native trimodality, and an Apache 2.0 license that actually respects developer freedom. Here’s why the 26B-A4B variant is about to eat your GPU.

#gemma#google-deepmind#mixture-of-experts...

gemma

Google’s TranslateGemma Proves That Size Doesn’t Matter: 12B Parameters Beat 27B

Google’s new TranslateGemma translation models challenge the ‘bigger is better’ orthodoxy, delivering superior performance with fewer parameters and running on consumer hardware.

#gemma#open-models#translation

function-calling

FunctionGemma’s Browser Physics Game is a Technical Marvel Marred by a Telling Demo

Google’s 270M-parameter function-calling model runs a physics puzzle game entirely in your browser, but early implementations expose the harsh gap between model capability and developer execution.

#function-calling#gemma#on-device-ai...

gemma

Gemma’s Multilingual Mirage: Why Google’s ‘Slow’ Release Cycle is Actually Winning

Debunking the myth that Google’s Gemma models are lagging behind, and exploring how their multilingual capabilities are quietly dominating the AI landscape.

#gemma#google-ai#llms...