Can I run Qwen3 8B on L40S?

Yes. Qwen3 8B runs on L40S (48 GB) at Q4_K_M — estimated 139 tok/s with 131K usable context.

Yes — it runs.best quant:

Decode speed

139 tok/sest

Usable context

131K

Memory

48 GB

Fits in memory139 tok/sest · 131K ctx

Fits in memory119 tok/sest · 131K ctx

Fits in memory79 tok/sest · 131K ctx