Skip to content

Can A100 40G run Qwen3-Coder 30B-A3B Instruct?

Comfortable: GGUF Q5_K_M needs 23.1 GB at 4K, batch 1; headroom 16.9 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_M19.8ComfortableVerified file
GGUF Q5_K_M23.1ComfortableVerified file
AWQ INT416.4ComfortableEstimated weights

Context capacity · GGUF Q5_K_M

ContextTotal GBVerdict
4K23.1Comfortable
16K24.4Comfortable
32K26.0Comfortable
128K35.9Tight

Generation speed

GGUF Q5_K_M: 767 tok/s · Bandwidth ceiling (not measured)

Other model / GPU pairs