Qwen3
Think Deeper or Act Faster
Start new thread
Zac Zuo

Qwen3.5 Small - 0.8B-9B native multimodal w/ more intelligence, less compute

by
Flowtica Scribe
•
Qwen just released the Qwen3.5 Small Model Series — 0.8B, 2B, 4B and 9B. Native multimodal with improved architecture and scaled RL. 0.8B and 2B are tiny and fast for edge devices, 4B makes a strong lightweight agent base, and 9B is already closing the gap with much larger models. Base versions released too.

Add a comment

Replies

Best
Jan Ahrend
I’ve been using qwen3:4B on my old gaming PC with n8n workflows for a year now. I just switched to qwen3.5:9B and (after setting parameters correctly) can say it works really well. I can now get performance that I’d only get on 32GB memory on 8GB memory.
Qaunain Meghjee

Nice, what is the amount of GPU or unified memory needed to run this model locally?

Shaurya Kesarwani

Soo surprised with 0.8b Model that could do fullscreen OCR aand object detection faster than most other models even on cpu taking roughly 10sec only and accuracy from my testing were spot on!

These open source models that are highly advanced yet small could be the way to go for the future of AI !

First
Previous
12