# 70-100 K Budget Recommendations — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/297a7bc3cbc11903

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-10-24 00:14 ingested 2026-10-24 02:40

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1xdudcy/70100_k_budget_recommendations/)

## retained text



```
Hey guys,

Assuming I have a budget of 70-100K USD budget.

What hardware would I get to be able to run GLM5.3 and Deepseek flash 1017 all for multiple concurrent users (10-20 or so) all doing agnetic tasks and long context of around 500k tokens each.

Also, is there anyway I can make the setup in a way that's eay/easy ish to be extendable sense models seem to be increasing in weights and size:)

Would love to hear your recommendations and suggestions
```



document fetched 2026-10-24 10:42 · canonical https://reddit.com/r/LocalLLaMA/comments/1xdudcy/70100_k_budget_recommendations
