Qwen2.5-7B-Instruct — performance reports

Community-submitted & scraped throughput measurements. Sortable by generation speed.

Ran this model? Log in to submit your own performance result.

Filter by GPU / rig

Sort by

Rig / GPU	Quant	Runtime	Context	Prompt tok/s	Gen tok/s	VRAM (GB)	Source	Reporter
2x RTX 3090	EXL2 · Q4_K_M	llama.cpp	4,096	3,791.45	168.49	39.48	user	Mr. Zachariah Bahringer MD