Post

Wrote up some notes on the new Qwen2.5-Coder-32B model, which is the first model I've run on my own Mac (64GB M2) that appears to be highly competent at writing code simonwillison.net/2024/Nov/12/...

simonwillison.netQwen2.5-Coder-32B is an LLM that can code well that runs on my MacThere’s a whole lot of buzz around the new Qwen2.5-Coder Series of open source (Apache 2.0 licensed) LLM releases from Alibaba’s Qwen research team. On first impression it looks like …

13/11/24, 2:31 am10 reposts94 likes

swyx

@swyx.io

13 Nov 2024

but is it worth the memory load vs just paying for tokens via api?

Simon Willison

@simonwillison.net

13 Nov 2024

Probably not, it's very hard to compete with the inexpensive hosted models these days

mr. TIM

@timkellogg.me

13 Nov 2024

can i also run slack?

Simon Willison

@simonwillison.net

13 Nov 2024

I also figured out how to run the Llama 3.2 Vision model in my Mac using LLM and Ollama and llm-ollama today simonwillison.net/2024/Nov/13/...

simonwillison.netOllama: Llama 3.2 VisionOllama released version 0.4 [last week](https://github.com/ollama/ollama/releases/tag/v0.4.0) with support for Meta's first Llama vision model, [Llama 3.2](https://ai.meta.com/blog/llama-3-2-connect-2...

Ann Catherine Jose

@annjose.com

14 Nov 2024

Thanks for the write up. I tried the Qwen model using ollama (M2 Mac 64 GB) just now. I asked it a bunch of questions that I had asked Perplexity for my current dev project, and it answered all of them well - including Next.js App router, Tailwind. Impressive results! Runs locally too! 🔥

Ann Catherine Jose

@annjose.com

14 Nov 2024

But all the answers start with "Certainly!" 😀 like the previous version of Claude. I guess it needs to be better aligned / tuned.

tani

@ewb.io

13 Nov 2024

does it have a vim plugin?