Head to head · October 2026
DGX Spark vs M5 Max 128GB: which is better for local AI?
128GB against 128GB, 273 against 614 GB/s. What each runs, how fast, and which one to buy.
The verdict
Both have 128GB of memory, so they run the same models. The M5 Max 128GB generates text about 85% faster thanks to 614 GB/s of memory bandwidth versus 273 GB/s. The M5 Max 128GB costs about $1,850 less ($5,100 vs $6,950).
- Get the DGX Spark if you care about power draw, heat, and noise or you want the widest software support (CUDA).
- Get the M5 Max 128GB if speed matters most or you want the most for your money.
We may earn a commission when you buy through links on this page. It never changes what we recommend. How we make money
Performance
Model by model
Estimated single-user generation speed for popular models, at the best quantization each card can hold with 8K context.
| Model | DGX Spark | M5 Max 128GB |
|---|---|---|
| Llama 3.1 8B Instruct | ~23 tok/sQ8_0 | ~44 tok/sQ8_0 |
| Qwen3 14B | ~13 tok/sQ8_0 | ~24 tok/sQ8_0 |
| gpt-oss-20b | ~43 tok/sMXFP4 | ~73 tok/sMXFP4 |
| Qwen3 30B-A3B (2507) | ~31 tok/sQ8_0 | ~54 tok/sQ8_0 |
| Gemma 3 27B | ~7.1 tok/sQ8_0 | ~14 tok/sQ8_0 |
| Qwen3 32B | ~5.8 tok/sQ8_0 | ~11 tok/sQ8_0 |
| Llama 3.3 70B Instruct | ~2.7 tok/sQ8_0 | ~5.3 tok/sQ8_0 |
| gpt-oss-120b | ~37 tok/sMXFP4 | ~64 tok/sMXFP4 |
Estimates from our bandwidth model; see methodology. Measured results, where we have them, are on each GPU's page.
Specifications
Spec by spec
| Spec | DGX Spark | M5 Max 128GB |
|---|---|---|
| Memory | 128GB LPDDR5X | 128GB LPDDR5X |
| Bandwidth | 273 GB/s | 614 GB/s● |
| Power | 240W● | 480W |
| Price | ~$6,950 | ~$5,100● |
| Price per GB | $54 | $40● |
| Architecture | Grace Blackwell (GB10) | Apple M5 Max |
| Released | 2025 | 2026 |
| Software | CUDA | Metal, MLX |
Our take
Bottom line on each
A quiet CUDA desktop with 128GB, but 273 GB/s limits speed and the price has climbed to about $6,950.
A quiet 128GB machine with more than twice the bandwidth of Strix Halo or DGX Spark.
Frequently asked questions
Is the DGX Spark or M5 Max 128GB better for running LLMs?
Both have 128GB of memory, so they run the same models. The M5 Max 128GB generates text about 85% faster thanks to 614 GB/s of memory bandwidth versus 273 GB/s. The M5 Max 128GB costs about $1,850 less ($5,100 vs $6,950).
How much faster is the M5 Max 128GB than the DGX Spark for AI?
About 85% faster at generating text in our estimates, because token generation is limited by memory bandwidth (614 GB/s vs 273 GB/s). Prompt processing depends more on compute and can differ by more.
Keep comparing