English
Deutsch
ไทย
Indonesia
Melayu
Việt Nam
FilipinoBuy the Box
Reserve with 25% today. Pay the remaining 75% only when your box ships within 30 days. Fully refundable before shipment — hardware is ready, software updates included.
.jpg&w=3840&q=75)
Mini PC Edge Node
Compact and fanless for small offices or single departments.
- Up to 5 concurrent users
- 8 GB VRAM / 32 GB RAM
- Silent, low-power, desktop form factor
AI Workstation
High-performance tower for power users and local AI R&D.
- Up to 20 concurrent users
- 24 GB VRAM / 64 GB RAM
- Dedicated GPU, expandable storage
Enterprise Rack Server
Datacenter-grade host for the whole organization.
- 50+ concurrent users
- Dual GPU / 128 GB RAM+
- Redundant power and ECC memory
Specifications & comparison
| Spec | API Lite Device for API only | Local AI Multimodal 100% local | Media Studio Full media generation + ComfyUI |
|---|---|---|---|
| GPU Architecture | Integrated | Ampere Next | Blackwell |
| AI Performance | Cloud-dependent | 200 TOPS | 500 TOPS |
| CUDA Cores | — | 4,096 | 8,192 |
| Memory | 8 GB LPDDR4 | 32 GB unified LPDDR5X | 64 GB unified LPDDR5X |
| Memory Bandwidth | — | 256 GB/s | 512 GB/s |
| Power (TDP) | 15 W | 40 W | 80 W |
| Cooling | Passive (fanless) | Active fan | Active + liquid ready |
| Storage | 128 GB eMMC | 1 TB NVMe SSD | 2 TB NVMe SSD |
| Networking | 1 GbE + Wi-Fi | 2.5 GbE | Dual 10 GbE |
| Dimensions | 120 × 80 × 30 mm | 180 × 120 × 45 mm | 210 × 150 × 55 mm |
| Weight | 0.4 kg | 1.2 kg | 2.1 kg |
| Warranty | 1 year | 2 years | 3 years |
| Price | $399 USD | $1,800 USD | $5,000 USD |
Running locally!
Why run locally?
- No recurring API subscriptions or token limits
- Your prompts and data never leave the device
- Low latency, no internet required for inference
- One-time hardware cost vs endless SaaS bills
What you can run
- Claude-style assistants via compatible local runners
- Vision, coding and chat models on-device
- Stable Diffusion and SeaDance 2.5 image pipelines
- Private RAG and agentic workflows
Orchestra OS
The operating system that ships out of the box with every hardware unit
ComfyUI via URL
Access ComfyUI's node-based image generation pipeline directly through a browser URL — no separate install needed.
Ollama + Local Models
Run multimodal models locally — text, vision, image, and coding models all on-device with Ollama pre-configured.
Remote Desktop UI
Full remote desktop access to the box from anywhere — manage your AI infrastructure with a graphical UI over the network.
Hermes Agent / Openclaw
Autonomous agent framework pre-installed — Hermes and Openclaw handle scheduled tasks, tool use, and agentic workflows out of the box.
OS Overhead & Features
Built-in system management keeps your box running reliably — whether on-site or remote.
Real-time system health and uptime dashboard — always know your box is operational.
Encrypted tunnel lets you access your box and all its services securely from anywhere — no VPN setup required.
Amortisation calculator
Compare what you currently spend on API calls each month against the one-time cost of the box.
Select option and check out
API Lite
Device for API only
AI: Cloud-dependent
Memory: 8 GB LPDDR4
Power: 15 W
Storage: 128 GB eMMC
Local AI
Multimodal 100% local
AI: 200 TOPS
Memory: 32 GB unified LPDDR5X
Power: 40 W
Storage: 1 TB NVMe SSD
Media Studio
Full media generation + ComfyUI
AI: 500 TOPS
Memory: 64 GB unified LPDDR5X
Power: 80 W
Storage: 2 TB NVMe SSD
.jpg&w=3840&q=75)
Order summary
25% deposit today. The remaining 75% is invoiced when your box ships. Fully refundable before shipment.