# I benchmarked Jev against 4 local LLMs on RTX5070 12GB - amazing. — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/ee61900690196d91

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-10-02 07:46 ingested 2026-10-03 02:41

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1wvns80/i_benchmarked_jev_against_4_local_llms_on_rtx5070/)

## retained text



```
[Updated Picture testing 5 more Agents. So in total its now 10 on my Rtx 5070 12GB](https://preview.redd.it/zl6v5mcn11th1.png?width=1315&format=png&auto=webp&s=09512311e73b58bfa5a3ad1479a7f0e3dd4dd675)

  
Here are the tests as you can see local Qwen 3.5 9B (Jebadiah 9B V2) is 4x lower latency on my rtx 5070 at same quality. 20 Tasks. Cheers.

Update:  
Changed the picture to feature 5 more models.
```



document fetched 2026-10-03 00:41 · canonical https://reddit.com/r/LocalLLaMA/comments/1wvns80/i_benchmarked_jev_against_4_local_llms_on_rtx5070
