Skip to content
VRAMwise

gpt-oss-20b (MoE): VRAM by quant

21.0B params · KV field None · max context field 131072.

QuantBitsWeight GB sketch
Q4_K_M4.85 bpw~13.0 GB
Q5_K_M5.7 bpw~15.1 GB
Q6_K6.6 bpw~17.4 GB
Q8_08.5 bpw~22.5 GB
MXFP44.25 bpw~11.6 GB
F16/BF1616.0 bpw~42.0 GB

Add KV for your context length in context VRAM. Overall fit: VRAM fit.

Educational. MoE active params may differ from total params.

Frequently asked questions

Size?

21.0B.

gpt-oss-20b (MoE) VRAM requirements — by VRAMwise ↗