# best <40B alternatives to Qwen/Deepseek for (1) Coding (2) Long document QA test — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/e5c9e4d653244de9

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-10-02 08:44 ingested 2026-10-03 02:41

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1wvondv/best_40b_alternatives_to_qwendeepseek_for_1/)

## retained text



```
Due to some reasons, Qwen/Deepseek Chinese models are NOT allowed in the my workplace. So, what local models, do you think, is the best alternatives to Qwen/Deepseek for

(1) **Coding**

(2) **Long document QA test** (like giving a long medical history of 120k tokens and ask a question based on that medical history)



**Gemma 31B ?** 

**Muse Glimmer 30B ?**  

**Nemotron ?**

also, I know that nothing beat qwen nowadays, but are there fine tunes from these alternative model that make them better than qwen3.8 27B  in terms of coding ?
```



document fetched 2026-10-03 00:41 · canonical https://reddit.com/r/LocalLLaMA/comments/1wvondv/best_40b_alternatives_to_qwendeepseek_for_1
