Compare configurations, memory capacity and hourly rates.
ASUS GX10 with 128 GB of unified memory
128 GB of unified memory for model weights, context and LLM experiments.
LLMs · code · images · video · finance
Explore this configurationTwo-node ASUS GX10 cluster
Run two workloads in parallel or distribute one model across both nodes.
LLMs · code · images · video · finance
Explore this configurationThe table lists the hourly rate and memory capacity for each configuration.
GX10 uses memory shared by the CPU and GPU. RTX and A100 use dedicated GPU memory.
The “from” rate applies to rentals of 169 hours or more.
The CPU, GPU, operating system and applications share LPDDR5x. The operating system and runtime use part of the memory.
Run independent workloads at the same time or distribute one model when your software supports multi-node execution.
| Configuration | Memory | Per hour |
|---|---|---|
| ASUS Ascent GX10NVIDIA GB10 · Blackwell | 128 GBCPU + GPU | Offer ends in—d—h—m—s |
| 2 × ASUS Ascent GX10NVIDIA GB10 · Blackwell | 2 × 128 GBunified per node | Offer ends in—d—h—m—s |
| RTX 3090 | 24 GBVRAM | $0.40/ hour |
| RTX 4090 | 24 GBVRAM | $0.712/ hour |
| RTX 5090 | 32 GBVRAM | $0.952/ hour |
| A100 SXM 80 GB | 80 GBVRAM | $1.432/ hour |
| H100 SXM 80 GB | 80 GBVRAM | $3.192/ hour |
Ampere GeForce with 24 GB of memory for CUDA workloads and GPU rendering.
Ada Lovelace GeForce for CUDA workloads, inference and GPU rendering.
Blackwell GeForce with 32 GB of memory for models and graphics workloads.
Professional RTX with ECC memory for AI, 3D and visualization.
96 GB of ECC memory on one GPU for AI and professional graphics.
A data center GPU with RT cores for inference, rendering and video workloads.
Ampere in the SXM form factor with 40 GB HBM2 for AI and scientific computing.
Ampere in the SXM form factor for training, inference and scientific computing.
Hopper in the SXM form factor with 80 GB HBM3 for data center AI workloads.
141 GB on one GPU for model weights, KV cache and working buffers.
Data center Blackwell with 180 GB HBM3e for memory-intensive workloads.
Two cards with 24 GB each for separate workloads or supported multi-GPU runs.
Four GPUs with 80 GB each for distributed workloads. Application support is required.
Eight GPUs with 80 GB each for distributed training and inference.
Check model memory, software compatibility and the cost of a trial run.
ASUS Ascent GX10 is a compact computer built on NVIDIA GB10 Grace Blackwell. Each node has 128 GB of memory shared by the CPU and GPU, and an ARM64 processor. Check that your containers and libraries support this architecture before renting.
Yes. Request one node or a pair. Both configurations use the same pool of two devices, so their availability is linked.
The catalog marks Qwen3-235B, Qwen3.5-397B and Llama 3.1 405B for the two-system ASUS GX10 cluster. Model cards list quantizations and weight sizes.
The catalog lists hourly rates. The ASUS GX10 starting rate applies to rentals of 169 hours or more; shorter rentals use a different rate. Availability and access dates are confirmed on request. The current form is a demo: it does not submit or store your information.
Yes. Submit a request on the GPU owners page with your hardware model, location and available hours. Rental terms and connectivity are arranged individually.