Google DeepMind's Gemma 4 update adds Flash Attention 4 support

Many users praised Gemma 4's Flash Attention 4 support and tool-use fixes from community feedback as meaningful improvements, while some called the changes insufficient and urged clearer naming like a 4.1 release.

Based on 30 visible X reactions from 69 accounts.

Digg Deeper

Ask a question below.

Published answers will appear here.

Commentary on X

Michał Pierzchała@thymikee

@googlegemma I heard @xeonbloomfield is already cooking in this area for Callstack Apex model 👀

12:31 PM · Jul 15, 2026

Greedisgood@Greedisgood2024

@googlegemma lol! Gemma … what is GDM doing??

12:26 PM · Jul 15, 2026

Martin Kemka@mkemka_

@osanseviero Thank you for the updates

1:18 PM · Jul 15, 2026

Tornado guy@fanofaliens

@googlegemma Wow Thanks Team 🫡

11:41 AM · Jul 15, 2026

View all

jeff kazzee@JeffKazzee

@googlegemma Lets go! Great job Google.

11:02 AM · Jul 15, 2026

Florian Brand@xeophon

@osanseviero OMG YEEEES

10:50 AM · Jul 15, 2026

Omar Sanseviero 🇫🇷 @RAISE Paris@osanseviero

We got tons of feedback about Gemma 4, so I'm excited to share a wave of improvements: - Flash Attention 4 support rolling out (significantly faster inference) - Patched bugs with tool use - Added resources to boost the vision capabilities with token budget management

10:48 AM · Jul 15, 2026

The release also fixes tool-use bugs and improves vision capabilities.