Qwen 3 0.6B & 8B First Look & LOCAL Testing (Reasoning + Coding)

Timestamps

00:00 - Intro

02:26 - 0.6B Python Testing

04:49 - 0.6B Second Python Test

05:33 - 0.6B HTML Test

06:58 - 0.6B HTML Improve Test

08:10 - 8B Python Game Test

10:07 - 8B Improved Game Test

11:05 - 8B HTML Test

12:01 - 8B HTML Improve Test

13:00 - Thoughts

14:04 - Odd Results From Early Release

14:49 - Closing Thoughts

In this video, we take a first look at two newly released models from the Qwen 3 family: the 0.6B model (tested at Q8 quantization) and the 8B variant (tested at Q4_K_M quantization). These models are part of the highly anticipated Qwen 3 release, and we explore their capabilities through local testing.

We start by running the 0.6B model through a few quick tasks, including Python game generation and basic HTML script creation. Following that, we switch over to the 8B model and run a similar set of tests, checking for improvements in output quality and reasoning.

This is a simple, hands-on overview of two promising models in the Qwen 3 lineup, with a focus on small-scale local inference performance and early practical observations from this release.

Hugging Face Model Repo: https://huggingface.co/Qwen