# Hilbert Super Workstation

[![Hilbert Super Workstation](./hilbert_super_workstation.jpg?thumbnail=400)](./hilbert_super_workstation.jpg)

[[OFFICIAL PRODUCT PAGE](https://hilbert.infplane.com/pages/hilbert)]
[[KICKSTARTER DESCRIPTION](https://www.kickstarter.com/projects/infplane/hilbert-agentic-computer/description)]

Infplane's Hilbert is a portable agentic supercomputer aimed at coding, agent workflows, local AI, CAD/EDA, and gaming. The launch messaging centers on an AMD Ryzen AI Max+ 395 platform, high-bandwidth local memory, and a compact 199 mm cube chassis.

Availability: crowdfunding / reserve-now launch.

### Highlights
- Designed for coding and creating agent workflows.
- Claims support for local 235B MoE or 8x32B LLMs.
- Promoted for OpenClaw, CAD/EDA, image generation, video generation, TTS, and AAA gaming.
- Mentions Windows, Cachy OS, SteamOS, and Fedora as supported operating systems.
- Highlights 96 GB "Ultimate VRAM" and 116 GB dynamic memory capacity.
- Claims 8000 MHz memory, with up to 8533 MHz configuration.
- Uses KLEENE, Infplane's motherboard-level control system, to coordinate CPU, GPU, and memory performance.

## Specifications

| Spec | Details |
|------|---------|
| **Dimensions** | 199 mm × 199 mm × 199 mm |
| **Operating System** | Windows 11 Pro (or as specified) |
| **Processor (CPU)** | AMD Ryzen AI Max + 395 Series |
| **Graphics (GPU)** | AMD Radeon™ 8060S Graphics |
| **System Memory (RAM)** | 128 GB LPDDR5X-8000 MHz (Up to 8533 MHz) |
| **Graphics Memory (VRAM)** | Up to 96 GB (Dynamic/Shared) |
| **Storage (Primary)** | 2 TB NVMe PCIe 4.0/5.0 SSD |
| **High-Speed I/O** | USB4 (with DisplayPort Alt Mode) × 2 |
| **Display Outputs** | DisplayPort 1.4 × 1, HDMI 2.1 × 1 |
| **Wired Networking** | 10 GbE × 1, 2.5 GbE × 1 |
| **Wireless Networking** | Wi-Fi 7, Bluetooth 5.4 |
| **Cooling System** | Multi-Island Full-Coverage Liquid Cooling |
| **Power Supply** | 1U 400W 80 Plus Platinum (92%+ Efficiency) |
| **Platform** | Infplane KLEENE control system (IMB1 intelligent motherboard controller) |

## Capabilities

- Supports 100B+ local AI models
- Enables 235B MoE or 8×32B LLM deployment
- OpenClaw integration for local inference ($0 API cost)
- Designed for agent workflows, CAD/EDA, image/video generation, TTS, and AAA gaming
- Supports Windows, CachyOS, SteamOS, and Fedora
- KLEENE enables dynamic power and performance coordination
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9