Gemma on NVIDIA Ada

Can I run Gemma 3 12B on an L40S?

Yes. Gemma 3 12B at Q4_K_M uses 8.6 GB of the 44.3 GB available on a L40S, and runs at about 67.5 tokens per second. You can push the context to 128K.

Runs comfortably

8.6 GB of 44.3 GB · 19%
044 GB
Weights 7.0 GB
KV cache 0.8 GB
Runtime overhead 0.8 GB

Gemma 3 12B at Q4_K_M leaves 35.7 GB spare on a L40S. There is room to raise the context length or move up a quantisation level.

Generation67.5tok/s
Prompt processing3116tok/s
Max context128Ktokens
KV per 1K tokens0GB

Every quantisation of Gemma 3 12B on a L40S

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

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 22.7 GB 24.4 GB Runs comfortably 128K 22.1
INT8 / W8A8 8.50 11.8 GB 13.5 GB Runs comfortably 128K 41.3
Q8_0 (GGUF) 8.50 11.8 GB 13.5 GB Runs comfortably 128K 41.3
FP8 (E4M3) 8.00 11.4 GB 13.0 GB Runs comfortably 128K 42.9
Q6_K 6.56 9.3 GB 10.9 GB Runs comfortably 128K 51.7
Q5_K_M 5.67 8.1 GB 9.7 GB Runs comfortably 128K 59.1
Q5_K_S 5.52 7.8 GB 9.5 GB Runs comfortably 128K 60.6
AWQ 4-bit 4.25 7.4 GB 9.0 GB Runs comfortably 128K 63.8
GPTQ 4-bit 4.25 7.4 GB 9.0 GB Runs comfortably 128K 63.8
MXFP4 4.25 7.4 GB 9.0 GB Runs comfortably 128K 63.8
Q4_K_M 4.85 7.0 GB 8.6 GB Runs comfortably 128K 67.5
Q4_K_S 4.58 6.6 GB 8.2 GB Runs comfortably 128K 70.7
Q4_0 4.55 6.6 GB 8.2 GB Runs comfortably 128K 71.1
IQ4_XS 4.25 6.2 GB 7.8 GB Runs comfortably 128K 75.2
Q3_K_M 3.91 5.7 GB 7.4 GB Runs comfortably 128K 80.3
IQ3_M 3.70 5.5 GB 7.1 GB Runs comfortably 128K 83.9
IQ3_XXS 3.06 4.6 GB 6.3 GB Runs comfortably 128K 97.0
Q2_K 2.63 4.1 GB 5.7 GB Runs comfortably 128K 108
IQ2_XXS 2.06 3.3 GB 5.0 GB Runs comfortably 128K 128
IQ1_M 1.75 2.9 GB 4.6 GB Runs comfortably 128K 142

Also worth checking

Gemma 3 12B on every GPU
Full hardware tableEverything that runs on a L40S
Full model tableWhy the KV cache is the part that bites
Guide