---
title: "Comment by Mihai_LeanZero on \"[Video] Running Qwen 3.8 27B Ternary Bonsai 2 on my 16GB Mac\" | Hashnode"
description: "Good that you're skeptical of the 98%-retention number. Those figures are almost always benchmark averages weighted toward knowledge/reasoning tests like MMLU,"
url: "https://hashnode.com/posts/video-running-qwen-3-8-27b-ternary-bonsai-2-on-my-16gb-mac/6ab09e2f4cb6159b7d516836/comment/6ab0a88bb1be516af34b7e80"
---

[Discussion](https://hashnode.com/posts/video-running-qwen-3-8-27b-ternary-bonsai-2-on-my-16gb-mac/6ab09e2f4cb6159b7d516836)

Comment [\[Video\] Running Qwen 3.8 27B Ternary Bonsai 2 on my 16GB Mac](https://hashnode.com/posts/video-running-qwen-3-8-27b-ternary-bonsai-2-on-my-16gb-mac/6ab09e2f4cb6159b7d516836)

[Mihai\_LeanZero](https://hashnode.com/@mihaileanzero)

[Sep 21](https://hashnode.com/posts/video-running-qwen-3-8-27b-ternary-bonsai-2-on-my-16gb-mac/6ab09e2f4cb6159b7d516836/comment/6ab0a88bb1be516af34b7e80)

Good that you're skeptical of the 98%-retention number. Those figures are almost always benchmark averages weighted toward knowledge/reasoning tests like MMLU, which measure something pretty different from a from-scratch generate-and-iterate task like the Three.js build. A model can score close to the original on one axis and still fail completely on the other, which is closer to what you actually hit over those three hours than a straight 2% quality gap would predict.

[Dragos Roua](https://hashnode.com/@dragosroua)

[Sep 21](https://hashnode.com/posts/video-running-qwen-3-8-27b-ternary-bonsai-2-on-my-16gb-mac/6ab09e2f4cb6159b7d516836/comment/6ab0a88bb1be516af34b7e80#reply-6ab0a9e906142ad8b2a099a7)

Yeah, I was pretty excited in the beginning, but after spending 3-4 hours on that Three.js generation task, I stepped back a little. Still, impressive.