Early Access
Reserve your HP-01.
Limited early-access units available. Enter your email to secure your HP-01 and start deploying AI locally.
Dedicated CPU
Agent Sandbox
- OpenClaw Pre-Installed, Running 24/7
- Layers of Encryption, Fully Protected
- Multi-User Supported, Up to 32 at Once
Dedicated GPU
Token Factory
- 0 Token Fee, 0 Maintenance Cost
- 300+ Local Models, Up to 235B Supported
- SOTA Agentic Capabilities, Full Data Privacy
Custom CUDA Kernels
Inference Engine
- TTFT: as low as 14ms
- Inference Speed: 42 Tokens/s
- Total Throughput: 1,010 Tokens/s
- TDP: 90W - 20x vs Ollama
SMB Team
OpenClaw for every team member.
32 concurrent users. One device. Zero IT overhead.
Solo Developer
Your personal AI lab.
300+ models at your fingertips. No cloud dependency.
Remote Access
OpenClaw Anytime, Anywhere.
Access your OpenClaw from anywhere with remote control while HP-01 hosts your OpenClaw 24/7.
Specifications
Performance
TTFT (Time to First Token)As Low as 14ms
Single-Request Speed42 Tokens/s
Total Token Throughput1,010 Tokens/s
Inference Capability (vs Ollama)20x Speedup
Hardware
Device Volume1.8 L
TDP90W
Noise36dB
Architecturex86 & arm & CUDA
Local Models Supported300+
Max Model Size235B Parameters
Software
Pre-installed ApplicationOpenClaw
Concurrent UsersUp to 32
Deployment Time5 Minutes for OpenClaw & 235B Model
APIOpenAI-Compatible Endpoint
Software ControlRemote Control Available