Heidenstedt 👩‍💻Jun 15

With my approach to AI-assisted coding, I currently spend about 5€ a month on API tokens.

I only use DeepSeek v4 Pro and local Qwen3.6-27B, so no subscriptions are needed.

If you actually engineer your code instead of just passively vibe coding, that is plenty.

#llm #ai #coding

Show threadmcl2k6Jun 15
@heidenstedt and what hardware do you use to run this?
Show threadHeidenstedt 👩‍💻

@mcl2k6

Sadly i am too poor to run DeepSeek v4 Pro localy with its 865 GB RAM and 200GB VRAM...

But Qwen3.6-27B is running on my NAS which has a Radeon RX 7900 XTX ... it has coil whine which makes for funny noises while inferencing.

Jun 15 at 12:56pmWeb
Show threadmcl2k6Jun 16
@heidenstedt Thank you very much for your answer. I want to try it out at some point, as I also have a discrete video card I bought recently, RX9070 (no XT) with 16GB of RAM.