Qwen on NVIDIA Ada

Can I run Qwen3 30B-A3B on an L40S?

Yes. Qwen3 30B-A3B at Q4_K_M uses 18.8 GB of the 44.3 GB available on a L40S, and runs at about 109 tokens per second. You can push the context to 128K.

Runs comfortably

18.8 GB of 44.3 GB · 42%
044 GB
Weights 17.3 GB
KV cache 0.8 GB
Runtime overhead 0.8 GB

Qwen3 30B-A3B at Q4_K_M leaves 25.5 GB spare on a L40S. There is room to raise the context length or move up a quantisation level.

Generation109tok/s
Prompt processing11380tok/s
Max context128Ktokens
KV per 1K tokens0GB

Every quantisation of Qwen3 30B-A3B on a L40S

Highlighted row is the highest quality that still fits at 8K context.

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 56.9 GB 58.4 GB Won't fit — 17.2
INT8 / W8A8 8.50 30.1 GB 31.6 GB Runs comfortably 128K 81.3
Q8_0 (GGUF) 8.50 30.1 GB 31.6 GB Runs comfortably 128K 81.3
FP8 (E4M3) 8.00 28.4 GB 30.0 GB Runs comfortably 128K 84.0
Q6_K 6.56 23.3 GB 24.9 GB Runs comfortably 128K 93.9
Q5_K_M 5.67 20.2 GB 21.7 GB Runs comfortably 128K 101
Q5_K_S 5.52 19.6 GB 21.2 GB Runs comfortably 128K 103
Q4_K_M 4.85 17.3 GB 18.8 GB Runs comfortably 128K 109
Q4_K_S 4.58 16.3 GB 17.9 GB Runs comfortably 128K 112
Q4_0 4.55 16.2 GB 17.8 GB Runs comfortably 128K 112
AWQ 4-bit 4.25 16.0 GB 17.5 GB Runs comfortably 128K 113
GPTQ 4-bit 4.25 16.0 GB 17.5 GB Runs comfortably 128K 113
MXFP4 4.25 16.0 GB 17.5 GB Runs comfortably 128K 113
IQ4_XS 4.25 15.2 GB 16.7 GB Runs comfortably 128K 116
Q3_K_M 3.91 14.0 GB 15.5 GB Runs comfortably 128K 120
IQ3_M 3.70 13.3 GB 14.8 GB Runs comfortably 128K 122
IQ3_XXS 3.06 11.1 GB 12.6 GB Runs comfortably 128K 131
Q2_K 2.63 9.6 GB 11.1 GB Runs comfortably 128K 138
IQ2_XXS 2.06 7.6 GB 9.1 GB Runs comfortably 128K 147
IQ1_M 1.75 6.5 GB 8.0 GB Runs comfortably 128K 153

Also worth checking

Qwen3 30B-A3B on every GPU
Full hardware tableEverything that runs on a L40S
Full model tableWhy the KV cache is the part that bites
Guide