ASUS Ascent GX10
ASUS's GB10 desktop AI system: 128 GB unified memory and the same DGX software as DGX Spark.
| AI performance | Up to 1 PFLOP FP4 |
|---|---|
| System memory | 128 GB LPDDR5x coherent unified memory |
| Superchip | NVIDIA GB10 Grace Blackwell |
| Networking | NVIDIA ConnectX-7 |
Your build
indicative — confirmed on quoteCheck a model against this machine before you buy it.
Pick the model and quantisation you plan to run. We'll show whether it fits in memory, how much room is left for context, and the throughput to expect. Estimates come from memory bandwidth and are replaced by our own bench figures where we have them.
Machine
Quantisation
Comfortable. 70 GB spare — room for 128k context, a second small model, or a RAG index in memory. Expect roughly 3.8 tok/s single-stream.
Machines that run this configuration
Weights 42.8 GB at 4.85 bits/param · KV cache 5.0 GB at 16k (fp16, approximate) · 10 GB reserved for OS and runtime. Speed is single-stream decode, memory-bound; prompt processing and batching behave differently. Or call an engineer on +44 7515 356 146.
A compact ASUS-built system on the same GB10 Grace Blackwell platform as DGX Spark, so models and containers move between them unchanged. A good fit for running and fine-tuning large language models locally, with no cloud bills and your data kept on site. Currently on backorder; we confirm the delivery date when you order.
Workstation prepares it before dispatch: DGX OS brought up to date, user accounts and remote access set up, and a starter model-serving stack (NIM, Ollama or vLLM) ready to try. We can link two units over their ConnectX-7 ports, run an onboarding session for your team and deal with UK warranty returns on your behalf. Support comes from our own UK engineers, and Trade members get free tech support on every order.