Confidential and open LLM inference for agents, paid per request in USDC
Stage
Established
Not given
Founded
Not given
Not given
Based in
US
US
Pricing model
Not given
Paid
Pricing
Not given
LLM inference: Models priced per million tokens input and output, ranging from $0.043 to $3.63 for input and $0.22 to $18.15 for output (e.g., Qwen3.8 27B route C: $0.043/$1.375, Kimi K3 route B: $3.63/$18.15). Virtual machines: Starter $0.159-$0.169/hour, Standard $0.318-$0.335/hour, Pro $0.636-$0.
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
Attested confidential model execution on Intel TDX hardware
Open tier LLM inference without attestation
Price shown before model execution
No account or API key required
Virtual machines for AI swarms by the hour
Phone numbers and calling with metered pricing
Email inboxes with sending allowances
Access via HTTP without signup
416 open source models available
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
Encrypted memory and disk in processor with no company access to keys
Price transparent and settled before request execution
No account signup, payment via wallet address
Automatic enforcement of calling and messaging consent rules
Agentgates