Skip to content

Can L40S 48G run K2 Horizon MoVA 36B-A4B?

Comfortable: GGUF Q6_K needs 33.0 GB at 4K, batch 1; headroom 15.0 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_M24.2ComfortableVerified file
GGUF Q5_K_M28.5ComfortableVerified file
GGUF Q6_K33.0ComfortableVerified file
GGUF Q8_042.5TightVerified file

Context capacity · GGUF Q6_K

ContextTotal GBVerdict
4K33.0Comfortable
16K35.5Comfortable
32K38.8Comfortable
128K58.5Does not fit

Generation speed

GGUF Q6_K: 277 tok/s · Bandwidth ceiling (not measured)

Other model / GPU pairs