what if you could instantly clone any object?
Gemma 4 is now on @Cerebras Inference, running up to 10x faster than GPUs (1,500 tokens/sec). Multimodal generations you can iterate on in real time :)
10:18 AM · Jun 22, 2026 · 7.4K Views
Sarah Chieng, Cerebras head of DevX, showcases Gemma 4 running at 1,500 tokens per second on Cerebras Inference
The demonstration generated editable 3D CAD models from images.
11816267.5K
Sentiment
Users are enthusiastic about Cerebras running Gemma 4 at 1,500 tokens per second for real-time multimodal 3D cloning because it promises impressive future capabilities like custom-designed weights.
Pos
100.0%
Neg
0.0%
3 comments with sentiment.
Cluster Engagement
Digg Deeper
No Digg Deeper questions have been answered for this story yet.
Posts from X
Most Activity
Most Activity
VIEWS411LIKES3
Philipp Schmid@_philschmid
@MilksandMatcha @cerebras Thats cool!
Sarah Chieng@MilksandMatcha
what if you could instantly clone any object?
Gemma 4 is now on @Cerebras Inference, running up to 10x faster than GPUs (1,500 tokens/sec). Multimodal generations you can iterate on in real time :)
5hViews 411Likes 3Bookmarks 0
Daniel Kim@learnwdaniel
@MilksandMatcha @cerebras ur gonna get so swol with your gemma designed weights!!
5hViews 59
AILOVER@ggg78g89
@MilksandMatcha @cerebras Great ,Sarah
6hViews 39