DGX Spark alternatives
Last updated
The main DGX Spark alternatives are the ASUS Ascent GX10 (same GB10 chip), 128 GB Ryzen AI Max+ 395 mini PCs, the Mac Studio and Olares One. Olares One trades capacity for a discrete 24 GB RTX 5090 Mobile and a ready server OS.
Olares One → Local AI server → Ollama · vLLM · Open WebUI → Private AI on your own network
What does DGX Spark offer?
DGX Spark offers 128 GB of unified memory on CUDA in a 150 x 150 x 50.5 mm box. NVIDIA lists 273 GB/s of memory bandwidth, a 20-core Arm CPU, up to 1 petaFLOP of FP4 compute and DGX OS. The 128 GB / 4 TB unit cost 6,950 USD on the NVIDIA store on 12 Nov 2026.
It weighs 1.2 kg and uses a 240 W power supply. Its strength is capacity on CUDA, since a 70B model needs about 40 GB at 4-bit by the approximate rule.
What are the best DGX Spark alternatives?
The closest alternative is the ASUS Ascent GX10, which uses the same GB10 chip and DGX OS with a 1 TB drive. The Ryzen AI Max+ 395 boxes match the 128 GB for less money and without CUDA. The Mac Studio and Olares One take different routes to local AI.
| Machine | Memory for models | CUDA | Price (read 12-13 Nov 2026) |
|---|---|---|---|
| NVIDIA DGX Spark | 128 GB unified | Yes | 6,950 USD (4 TB) |
| ASUS Ascent GX10 | 128 GB unified (same GB10) | Yes | 5,999 USD (1 TB; out of stock on 12 Nov) |
| Minisforum MS-S1 Max | 128 GB shared | No | 3,959 USD (sale) |
| GMKtec EVO-X2 | 128 GB shared | No | 3,499.99 USD (sale) |
| Mac Studio (M5 Ultra) | 96 GB unified, base | No | 5,499 USD |
| Olares One | 24 GB VRAM plus 96 GB RAM | Yes | 4,999 USD list |
When is Olares One the better choice?
Olares One is the better choice when your models stay at or below about 32B and several people use them. It has 24 GB of dedicated GDDR7 on an RTX 5090 Mobile and an x86 CPU, so packages without Arm builds install normally, and Olares OS adds accounts, remote access and backups.
In Olares' own test, Qwen3-30B-A3B ran at 157 tokens/s for one request and about 81 tokens/s per request with 8 concurrent requests. No side-by-side test with DGX Spark is in our data.
When should you still buy DGX Spark?
Buy DGX Spark or the GX10 if you run 70B models or larger every day, or if you build for NVIDIA's own software stack. Olares One cannot hold a dense 70B model in its 24 GB of VRAM, and at a 4,999 USD list price it is no cheap route to large-model capacity.
It can run bigger models from system memory, slowly. Olares' FAQ gives about 36 tokens/s for gpt-oss-120b, with the part that does not fit in VRAM running on the CPU. It also has a single 2.5 GbE port. See the hardware guide, the GPU guide and Mac Studio alternatives.
Questions & answers
Is the ASUS Ascent GX10 the same as DGX Spark?
It uses the same NVIDIA GB10 chip, 128 GB of unified memory and NVIDIA DGX OS. The priced unit has a 1 TB drive at 5,999 USD on the NVIDIA store, and it was out of stock on 12 Nov 2026.
Is Olares One faster than DGX Spark?
No side-by-side test is in our data, so we make no claim. Olares One has a discrete GPU with dedicated VRAM, and DGX Spark has more memory at 273 GB/s according to NVIDIA. For a model that fits in 24 GB, compare published tokens/s for that exact model.
Sources for this entry
- NVIDIA DGX Spark: maker's page (Maker-published)
- ASUS Ascent GX10: maker's page (Maker-published)
- Minisforum MS-S1 Max: maker's page (Maker-published)
- GMKtec EVO-X2 (Ryzen AI Max+ 395, 128GB): maker's page (Maker-published)
- Apple Mac Studio (M5 Ultra): maker's page (Maker-published)
- Olares One FAQ (Olares docs) (Company-published)
- Olares blog: local AI hardware benchmarking (Oct 2025) (Company-published)
Continue this reading path
How the related websites connect
Cite this entry
Best Local AI Server. “DGX Spark alternatives.” Content reviewed 13 November 2026.
https://bestlocalaiserver.com/dgx-spark-alternative/