Home / Will it run? / gpt-oss-20b
OpenAI · gpt-oss
Will gpt-oss-20b run on my GPU?
21B parameters (mixture of experts, 3.6B active per token, but every expert must sit in memory), 131,072-token context, Apache-2.0. At Q4_K_M the weights come to about 11.7 GiB; add 1.5 GiB for the runtime and you need roughly 13.2 GiB of VRAM before context.
Weights by quant
GiBQ4_K_M · the everyday quant11.7 GiB
Q8_0 · near-lossless20.7 GiB
FP16 · full weights39.1 GiB
Runtime reserve1.5 GiB
Card by card
11 of 18 fit at Q4Headroom is what remains for the KV cache after weights and the 1.5 GiB reserve. How far it stretches depends on the architecture: grouped-query models are frugal, older dense models are not. If the table says tight, plan on a shorter context or a lower quant. Parameter count from the model's own card. The Buy column opens live Amazon listings; those links may earn us a commission.