OpenWeightsTerminal

Pair

MiMo-V2.6-Distill-Qwen-9B

RTX 5060 8GB · Q5_K_M · llama.cpp

Decode · tok/s

47

Prefill

1600

tok/100W

32.4

full 262K context. ~47 tok/s decode, ~1600 tok/s prefill.

Qwen3.5-9B fine-tuned on MiMo-V2.6 outputs. 8 GB card, not the 5060 Ti.

Officialclaimed