The export script was the bug: first LLM numbers on QCS6490 (Qwen3-0.6B, 1,555 tok/s prefill)

Three days ago I asked a simple question: how fast does a small LLM run on a Qualcomm QCS6490?...

Read Original

Related