Local AI workstation for private inference with up to 192 GB of unified memory and 32 GB Radeon GPU
Stage
Established
Not given
Founded
Not given
Not given
Based in
US
Not given
Pricing model
Not given
Paid
Pricing
Not given
Developer: starting at $5,999 USD (128 GB memory, 2 TB storage). Memory upgrade to 192 GB adds $2,200. Storage upgrade to 4 TB adds $250. Cluster kit $799 per machine. Business pricing available by quote. Worldwide shipping, import duties and taxes outside US included. No sales tax in US.
Key features
GroqMetal: dedicated bare-metal infrastructure tuned for speed and reliability
GroqCore: production-ready inference stack with no infrastructure expertise required
GroqAssured: enterprise-grade governance, auditability, and control
256 LPUs per rack with 40 PB/s SRAM bandwidth
1,000 tokens per second per user throughput
128 GB on-chip SRAM per rack
315 PFLOPS of FP8 inference compute
13 globally distributed data centers across four continents
Run DeepSeek V4 Flash and Qwen 27B models
Browser-based setup over encrypted Bluetooth
Private OpenAI-compatible API
Lucebox Mix: run two models simultaneously
Headless Ubuntu Server with pre-installed inference engine
Optional Tailscale integration for remote access
Optional cluster kit with 100 GbE networking
What makes it different
Fast inference at scale without sacrificing affordability
LPU pioneered technology integrated with NVIDIA GPUs
Fully integrated platform combining infrastructure, inference, and control
No per-token billing or cloud dependencies
1.5x faster decode speed than DGX Spark
Hardware and software co-designed for optimal performance
Completely private: runs offline with no cloud proxy