Skip to content

Can L40S 48G run Seed-OSS 36B Instruct?

Comfortable: GGUF Q5_K_M needs 27.8 GB at 4K, batch 1; headroom 20.2 GB.

Text input, FP16 KV cache, one card. Unified-memory capacity uses the GPU guide’s usable limit; negative headroom may require host offload. Speed source settings may differ from this 4K estimate.

Measured runs

No measured run for this pair. Sizing is an estimate, not a runtime test.

Quant memory at 4K

QuantTotal GBVerdictWeight sizing
GGUF Q4_K_M23.9ComfortableVerified file
GGUF Q3_K_M19.5ComfortableVerified file
GGUF Q5_K_M27.8ComfortableVerified file
AWQ INT423.2ComfortableVerified file

Context capacity · GGUF Q5_K_M

ContextTotal GBVerdict
4K27.8Comfortable
16K31.1Comfortable
32K35.5Comfortable
128K61.9Does not fit

Generation speed

GGUF Q5_K_M: 36 tok/s · Bandwidth ceiling (not measured)

Other model / GPU pairs