EDGE OSAI HARDWARE COMPARISON
DGX Spark vs Ryzen AI Max+ 395Specs, memory and LLM performance
Both are compact desktops with 128 GB of unified LPDDR5X and similar bandwidth (273 GB/s vs 256 GB/s), so they generate tokens at similar speeds. DGX Spark adds NVIDIA’s CUDA software and far more AI compute; Ryzen AI Max+ 395 mini PCs cost about half as much.
Memory
128 GB vs 128 GB
Bandwidth
273 GB/s vs 256 GB/s
Power
240 W power supply vs 45 to 120 W
Type
Desktop AI system vs Mini PC chip
Which should you choose?
CHOOSE DGX SPARK IF
You want CUDA and NVIDIA’s AI software, faster prompt processing and fine-tuning, or to link two units for 256 GB.
CHOOSE RYZEN AI MAX+ 395 IF
You want the lowest-cost way to run 70B to 120B models locally with llama.cpp, LM Studio or ROCm.
Still deciding between DGX Spark and Ryzen AI Max+ 395?
Tell us your models, users and data. We will recommend a setup and run it inside your own building with Vanik Edge OS. Free, no obligation.
DGX Spark vs Ryzen AI Max+ 395 specs
Manufacturer specifications. Green marks the stronger figure where a higher number is better. Dense compute is shown without sparsity.
| Specification | NVIDIA DGX Spark | AMD Ryzen AI Max+ 395 (Strix Halo) |
|---|---|---|
| Type | Desktop AI system | Mini PC chip (GMKtec EVO-X2, Framework Desktop and others) |
| Memory | 128 GB | 128 GB |
| Memory type | LPDDR5X unified memory | LPDDR5X-8000 unified memory (up to 96 GB as VRAM on Windows, more on Linux) |
| Memory bandwidth | 273 GB/s | 256 GB/s |
| FP16 / BF16 (dense) | Not published | Not published |
| FP8 (dense) | Not published | Not published |
| FP4 (dense) | 500 TFLOPS | Not published |
| Power | 240 W power supply | 45 to 120 W |
| Multi-GPU link | ConnectX-7 200 Gb/s: link two units for 256 GB | USB4 and 2.5–10 GbE, depending on the system |
| Form factor | Palm-sized desktop with 4 TB NVMe | Mini PCs and small desktops |
| Architecture | GB10 Grace Blackwell, 2025 | Zen 5 + RDNA 3.5, 2025 |
| Software | CUDA, NVIDIA AI software, vLLM, llama.cpp, Ollama | ROCm, Vulkan, llama.cpp, LM Studio, Ollama |
| Price | US$3,999 at launch (4 TB) | From about US$2,000 for 128 GB systems |
What they mean for running LLMs
For large language models, memory decides which models fit and how many users you can serve, and memory bandwidth decides how fast each user gets tokens. These are planning figures for one GPU or system using a 4-bit model.
| For LLMs | DGX Spark | Ryzen AI Max+ 395 |
|---|---|---|
| Usable for models | About 109 GB | About 86 GB (up to 96 GB can be set as VRAM on Windows) |
| Largest dense model at 4-bit (1 user, short context) | About 170B parameters | About 140B parameters |
| Llama 3.3 70B at 4-bit, single-user speed ceiling | About 7 tokens/s | About 7 tokens/s |
Which open-weight models fit
Memory needed for one user with 8K context at the lowest precision available for each model. Open a model for team-size and fine-tuning figures.
| Model (1 user, 8K context, lowest available precision) | DGX Spark | Ryzen AI Max+ 395 |
|---|---|---|
| gpt-oss-20b 21B · about 15 GB at MXFP4 | Yes | Yes |
| Gemma 4 26B A4B 25.2B · about 16 GB at 4-bit | Yes | Yes |
| Qwen3.6-27B 27.78B · about 20 GB at 4-bit | Yes | Yes |
| Qwen3.8-27B 27.8B · about 24 GB at NVFP4 | Yes | Yes |
| Gemma 4 31B 30.7B · about 20 GB at 4-bit | Yes | Yes |
| Qwen3.6-35B-A3B 35B · about 24 GB at NVFP4 | Yes | Yes |
| Llama 3.3 70B Instruct 70.6B · about 46 GB at INT4 | Yes | Yes |
| gpt-oss-120b 117B · about 70 GB at MXFP4 | Yes | Yes |
| MiniMax-M2.5 229B · about 135 GB at 4-bit | 2 linked units | No |
| GLM-5.3-Flash 320B · about 192 GB at NVFP4 | 2 linked units | No |
| DeepSeek-V4.1-Flash 552B · about 320 GB at 4-bit | No | No |
| DeepSeek-R1 671B · about 410 GB at 4-bit | No | No |
| DeepSeek-V3.2 685B · about 419 GB at NVFP4 | No | No |
| GLM-5.2 744B · about 455 GB at NVFP4 | No | No |
| GLM-5.3 753B · about 460 GB at NVFP4 | No | No |
| GLM-5.1 754B · about 461 GB at NVFP4 | No | No |
| Kimi K2 1T · about 579 GB at 4-bit | No | No |
| DeepSeek-V4-Pro 1.6T · about 976 GB at FP4 + FP8 | No | No |
| Kimi K3 2.8T · about 1.62 TB at MXFP4 | No | No |
Models in amber need several of these GPUs working as one. Buying the GPUs is the easy part. Vanik Edge OS splits the model across them, keeps it running if a server fails and manages it all from one console. See how Edge OS clusters GPU servers
FROM SIZING TO RUNNING
Sizing the hardware is the easy part.
Getting several GPUs to work as one system for a whole company is where most projects slow down. That is the part Vanik Edge OS does for you.
STEP 01 · THIS TOOL
Size it
Work out the GPU memory your models need and which GPUs or servers can hold them.
STEP 02 · VANIK EDGE OS
Cluster it
Edge OS joins your GPU servers into one cluster, splits a large model across them and keeps it running if a server fails. NVIDIA and AMD servers work side by side.
STEP 03 · VANIK EDGE OS
Run it
Your team gets AI apps with single sign-on and an audit trail, managed from one console, with your data staying inside your building.
DGX Spark vs Ryzen AI Max+ 395 questions
What is the difference between the DGX Spark and the Ryzen AI Max+ 395?
Both are compact desktops with 128 GB of unified LPDDR5X and similar bandwidth (273 GB/s vs 256 GB/s), so they generate tokens at similar speeds. DGX Spark adds NVIDIA’s CUDA software and far more AI compute; Ryzen AI Max+ 395 mini PCs cost about half as much.
DGX Spark vs Ryzen AI Max+ 395: which is better for LLM inference?
For generating tokens, the DGX Spark is faster: it has 273 GB/s of memory bandwidth versus 256 GB/s, and LLM token generation is mostly limited by bandwidth. Choose the DGX Spark if you want CUDA and NVIDIA’s AI software, faster prompt processing and fine-tuning, or to link two units for 256 GB. Choose the Ryzen AI Max+ 395 if you want the lowest-cost way to run 70B to 120B models locally with llama.cpp, LM Studio or ROCm.
Can the DGX Spark and Ryzen AI Max+ 395 run Llama 3.3 70B?
Llama 3.3 70B needs about 46 GB of memory at 4-bit for one user with 8K context. The DGX Spark runs it on one unit at about 7 tokens/s at most for one user, and the Ryzen AI Max+ 395 runs it on one unit at about 7 tokens/s at most for one user. Speeds are theoretical ceilings from memory bandwidth; real speeds are usually 50 to 80% of that.
How much does the DGX Spark cost compared with the Ryzen AI Max+ 395?
DGX Spark: US$3,999 at launch (4 TB). Ryzen AI Max+ 395: From about US$2,000 for 128 GB systems. Data center GPUs are usually bought as part of 4- or 8-GPU servers, so compare full-system quotes.
How are these numbers calculated?
Memory, bandwidth, compute and power are manufacturer specifications. Model memory needs come from our VRAM calculator: weights plus KV cache plus about 5 to 10% runtime overhead, keeping 10% of each GPU free (15% on GB10 systems, 25% on Mac). The speed ceiling is memory bandwidth divided by the model’s weight size. All figures are planning estimates.
Related comparisons
Specifications from the manufacturer pages for the NVIDIA DGX Spark and AMD Ryzen AI Max+ 395. Speed ceilings are bandwidth divided by model size, before batching; real throughput depends on the inference engine and settings. Last checked November 10, 2026.
Not sure which hardware fits your team?
Vanik Edge OS runs open-weight models on NVIDIA DGX, AMD Instinct and other GPU servers on your premises, and helps you size the hardware first.