NVIDIA DGX Spark vs Mac Studio M5 Max 40-core GPU
Prices, stock and measured speed side by side. Last checked 2026-10-16.
Verdict
NVIDIA DGX Spark is cheaper in USD by $400 (9%). In EUR, NVIDIA DGX Spark is cheaper by €1,733 (at least one EUR price is converted). On gpt-oss-120b, Mac Studio M5 Max 40-core GPU is faster (79.1 vs 58.72 tok/s; sources and engines differ, see below). Per dollar, Mac Studio M5 Max 40-core GPU gives more large-model speed (15.5 vs 12.5 tok/s per $1k).
Generated from the numbers below; no hands-on testing.
| NVIDIA DGX Spark (Founders Edition) | Mac Studio M5 Max 40-core GPU, 128GB / 512GB | |
|---|---|---|
| Price (USD) | $4,699 | $5,099 |
| Price (EUR) | ≈€4,176 | €5,909 |
| Memory | 128GB | 128GB |
| Stock | not tracked shop | not tracked shop |
| gpt-oss-120b tok/s | 58.72 | 79.1 |
| gpt-oss-120b tok/s per $1k | 12.5 | 15.5 |
| gpt-oss-120b tok/s per €1k | 14.1 | 13.4 |
| Llama-2 7B tok/s | 57.21 | 119.08 |
| Llama-2 7B tok/s per $1k | 12.2 | 23.4 |
| Llama-2 7B tok/s per €1k | 13.7 | 20.2 |
| Stores | US store · EU store | US store · EU (Ireland) |
Benchmark sources
- NVIDIA DGX Spark: large-model source (llama.cpp CUDA, llama-bench tg32, b7946) · small-model source
- Mac Studio M5 Max 40-core GPU: large-model source (Ollama 0.34.1 (not llama-bench)) · small-model source
Large: gpt-oss-120b MXFP4 (59 GiB weights), text generation tok/s at empty context, single request. Small: llama.cpp llama-bench, Llama-2 7B Q4_0, tg128 (text generation tok/s); best of FA on/off as posted.