DGX Spark
21 records across 8 models (quants: BF16, FP8, NVFP4, Q5_K_M, Q6_K, i1-Q3_K_M, unspecified).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DGX Spark | CYBER-FROST-3.8-NVFP4 | NVFP4 | vLLM | 155.6 | – | – | – | – | 2026-10-01 | community | – | source |
| DGX Spark | CYBER-FROST-3.8-NVFP4 | NVFP4 | vLLM | 137.6 | – | – | – | – | 2026-10-01 | community | – | source |
| DGX Spark | EXAONE-4.5-33B | FP8 | vLLM | 5.57 | – | – | – | – | 2026-10-07 | community | – | source |
| DGX Spark | Gemma-31B | Q5_K_M | llama.cpp (Vulkan) | – | 646.7 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Gemma-31B | Q5_K_M | llama.cpp (Vulkan) | – | 726.2 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Gemma-31B | Q5_K_M | llama.cpp (Vulkan) | – | 636.9 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Gemma-31B | Q5_K_M | llama.cpp (Vulkan) | – | 745.1 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Llama 3.1 8B | unspecified | vLLM | 15.83 | – | – | – | – | 2026-10-07 | community | – | source |
| DGX Spark | Llama 3.1 8B | Q6_K | llama.cpp (Vulkan) | – | 2238.9 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Llama 3.1 8B | Q6_K | llama.cpp (Vulkan) | – | 2419.3 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Llama 3.1 8B | Q6_K | llama.cpp (Vulkan) | – | 2316.6 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Llama 3.1 8B | Q6_K | llama.cpp (Vulkan) | – | 2688.4 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Qwen3.5 35B | unspecified | llama.cpp (CUDA) | 63.5 | – | – | – | tg128 | 2026-04-01 | community | – | source |
| DGX Spark | Qwen3.5 35B | unspecified | llama.cpp (CUDA) | 50.1 | – | – | – | tg128 | 2026-04-01 | community | – | source |
| DGX Spark | Qwen3.5-27B | i1-Q3_K_M | llama.cpp (CUDA) | – | 94 | – | – | – | 2026-04-02 | community | – | source |
| DGX Spark | Qwen3.5-27B | i1-Q3_K_M | llama.cpp (CUDA) | – | 80 | – | – | – | 2026-04-02 | community | – | source |
| DGX Spark | Qwen3.5-35B-A3B 35B MoE A3B | BF16 | vLLM | 431.42 | – | – | – | – | 2026-02 | community | – | source |
| DGX Spark | Qwen3.8-27B | Q6_K | llama.cpp (Vulkan) | – | 605.0 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Qwen3.8-27B | Q6_K | llama.cpp (Vulkan) | – | 693.8 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Qwen3.8-27B | Q6_K | llama.cpp (Vulkan) | – | 602.8 | – | – | – | 2026-10 | community | – | source |
| DGX Spark | Qwen3.8-27B | Q6_K | llama.cpp (Vulkan) | – | 731.5 | – | – | – | 2026-10 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085