Tired of hitting token limits on paid AI models like ChatGPT, Claude, or Gemini? In this video, I test Qwen 3.5 — a free, open-source language model you can run locally on your own machine — to see if it can work as a real alternative.

I'll show you how to install and run Qwen 3.5 using Ollama and LM Studio, then put it through real coding tasks to find out where it shines and where it falls short.

The Biggest Model Result (Sorting Algorithm Visualizer): https://blokhin.us/sorting.html

--------

CONNECT WITH ME
📱 Twitter / X: https://twitter.com/Zero_to_MVP
👨‍💻 Linkedin: https://www.linkedin.com/in/nblokhin/  

--------

⏱️ Timestamps:
00:00 Intro
01:33 How to Run Models Locally
03:50 First Test
05:08 Second Test
06:53 Testing Biggest Model (397B Parameters)
09:06 Conclusions

👍 If you found this helpful, please like and subscribe for more dev tools and programming content!

YouTube

A hands-on evaluation of running Qwen 3.5 locally using Ollama and LM Studio as a free alternative to paid AI models like Claude, ChatGPT, and Gemini. The 9B parameter local version struggled with complex code generation tasks but handled small, well-defined tasks well. The 397B cloud-hosted version via Ollama performed comparably to paid models, successfully completing a sorting algorithm visualizer from a spec. Key takeaway: model size matters significantly, and while small local models have limited use cases, the largest Qwen 3.5 models can compete with commercial offerings.

Qwen 3.5 Locally: Can It Replace Paid AI Models?