DeepSeek-V3: Has anyone run it locally and how was the experience

#8
by Daniyal643 - opened

I’m currently using local models like Qwen2.5-Coder with Ollama in VS Code, and I’m curious about DeepSeek-V3.

Has anyone here actually tried running it locally or in a lightweight setup?

What hardware did you use (CPU/GPU/VRAM)?
What inference tool worked best (Ollama, vLLM, etc.)?
Was it usable for coding tasks or too heavy?
How does it compare to smaller models like Qwen2.5-Coder?

Sign up or log in to comment