Skip to content

Can RTX 5090 run Qwen3.8 27B?

Comfortable: GGUF Q6_K needs 25.2 GB at 4K, batch 1; headroom 6.8 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_M18.5ComfortableVerified file
GGUF Q3_K_M14.3ComfortableVerified file
GGUF Q5_K_M22.2ComfortableVerified file
GGUF Q6_K25.2ComfortableVerified file
GGUF Q8_030.7TightVerified file

Context capacity · GGUF Q6_K

ContextTotal GBVerdict
4K25.2Comfortable
16K26.0Comfortable
32K27.1Comfortable
128K33.7Does not fit

Generation speed

GGUF Q6_K: 46–61 tok/s · Derived range

Other model / GPU pairs