Skip to content

Can RTX 4090 run Qwen3 30B-A3B Instruct?

Comfortable: GGUF Q4_K_M needs 19.7 GB at 4K, batch 1; headroom 4.3 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

FrameworkQuanttok/sVRAM used (GB)
llama.cppGGUF Q4_K_M9519

Quant memory at 4K

QuantTotal GBVerdictWeight sizing
GGUF Q4_K_M19.7ComfortableEstimated weights
GGUF Q5_K_M23.0TightEstimated weights
AWQ INT416.4ComfortableEstimated weights

Context capacity · GGUF Q4_K_M

ContextTotal GBVerdict
4K19.7Comfortable
16K21.0Comfortable
32K22.6Tight

Generation speed

Measured: 95 tok/s, GGUF Q4_K_M, llama.cpp

GGUF Q4_K_M: 95 tok/s · Measured

Other model / GPU pairs