RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can gemma 3 4b pt run on the DGX Spark?

Yes. At 8K context the best fit is F16, using about 9.2 GB of the 109 GB usable on the DGX Spark. Estimated speed ceiling: ~32 tokens/sec.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K)2.6 GB Fits2.7 GB Fits3.1 GB Fits5.0 GB Fits169
3-bit (Q3_K_M)3.0 GB Fits3.1 GB Fits3.6 GB Fits5.5 GB Fits130
4-bit (Q4_K_M)3.5 GB Fits3.6 GB Fits4.1 GB Fits6.0 GB Fits104
5-bit (Q5_K_M)3.9 GB Fits4.0 GB Fits4.5 GB Fits6.4 GB Fits89
6-bit (Q6_K)4.4 GB Fits4.5 GB Fits4.9 GB Fits6.8 GB Fits77
8-bit (Q8_0)5.3 GB Fits5.4 GB Fits5.9 GB Fits7.8 GB Fits60
16-bit (F16/BF16)9.1 GB Fits9.2 GB Fits9.6 GB Fits11.5 GB Fits32

• real GGUF file size. Speed is a bandwidth ceiling (273 GB/s), not a benchmark. Methodology.

More

gemma 3 4b pt on every GPU and Mac · Every model that runs on the DGX Spark · Other models that fit 64 GB VRAM · Try other settings in the calculator

Data: Hugging Face Hub (google/gemma-3-4b-pt), NVIDIA specs. Data as of 2026-11-17.