Skip to content

Can RTX 4090 run Seed-OSS 36B Instruct?

Comfortable: GGUF Q3_K_M needs 19.5 GB at 4K, batch 1; headroom 4.5 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_M23.9TightVerified file
GGUF Q3_K_M19.5ComfortableVerified file
GGUF Q5_K_M27.8Does not fitVerified file
AWQ INT423.2TightVerified file

Context capacity · GGUF Q3_K_M

ContextTotal GBVerdict
4K19.5Comfortable
16K22.8Tight
32K27.2Does not fit
128K53.6Does not fit

Generation speed

GGUF Q3_K_M: 60 tok/s · Bandwidth ceiling (not measured)

Other model / GPU pairs