RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can Kimi VL A3B Thinking 2506 run on the DGX Spark?

Yes. At 8K context the best fit is F16, using about 60.4 GB of the 109 GB usable on the DGX Spark. Estimated speed ceiling: ~23 tokens/sec.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K)6.6 GB Fits6.7 GB Fits7.4 GB Fits10.2 GB Fits243
3-bit (Q3_K_M)8.3 GB Fits8.4 GB Fits9.1 GB Fits12.0 GB Fits186
4-bit (Q4_K_M) •10.6 GB Fits10.8 GB Fits11.5 GB Fits14.3 GB Fits142
5-bit (Q5_K_M)11.7 GB Fits11.8 GB Fits12.5 GB Fits15.4 GB Fits128
6-bit (Q6_K)13.4 GB Fits13.5 GB Fits14.2 GB Fits17.0 GB Fits111
8-bit (Q8_0) •16.6 GB Fits16.8 GB Fits17.5 GB Fits20.3 GB Fits88
16-bit (F16/BF16) •60.3 GB Fits60.4 GB Fits61.1 GB Fits64.0 GB Fits23

• real GGUF file size from ggml-org/Kimi-VL-A3B-Thinking-2506-GGUF. Speed is a bandwidth ceiling (273 GB/s), not a benchmark. Methodology.

More

Kimi VL A3B Thinking 2506 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 (moonshotai/Kimi-VL-A3B-Thinking-2506), NVIDIA specs. Data as of 2026-11-17.