AI Devices & Configurations

The reference GB10 desktop AI appliance with ~128GB unified memory.

ASUS GB10-class mini AI appliance for desktop-side local inference.

Acer GB10-class AI mini system (expected class).

Dell GB10-class desktop AI system (expected class).

Lenovo GB10-class compact AI station (expected class).

Reference small-office appliance: a private assistant on a quiet box.

Compact, quiet Apple-silicon desktop for entry local AI.

More bandwidth and memory for mid-size local models in a small box.

Quiet mini PC with large unified memory for sizable models.

Compact, repairable Ryzen AI Max desktop with unified memory.

High unified memory and bandwidth — runs 70B models quietly.

Very large unified memory for flagship local models on a desktop.

High-end desktop usable as a capable single-GPU AI box.

Expandable Apple tower with up to ~192GB unified memory.

Creator workstation with a 24GB RTX 4090 for local AI.

Creator workstation tuned for GPU-accelerated AI work.

Professional workstation with a 48GB pro GPU — 70B models on one board.

Heavy-duty Z workstation with a 48GB pro GPU for serious agents.

Dual-socket workstation chassis with a 24GB RTX 4090.

Local-AI workstation; pool two 24GB cards for bigger models.

Reference multi-GPU workstation for a private Business Command Center.

Tower AI workstation with 48GB datacenter cards.

8-way H100 server for on-prem AI hosting at scale.

8-way H200 server with large HBM for the biggest private models.

8-way MI300X server — class-leading per-accelerator memory.

Rentable H100 instances; pricing varies — verify current rates.

Rentable H100 instances aimed at AI — verify current rates.

Rentable L40S instances for cost-effective inference — verify rates.