Agent-native AI infrastructure for building, running and scaling AI from models to agents
Stage
Established
Not given
Founded
Not given
Not given
Based in
US
FR
Pricing model
Not given
Freemium
Pricing
Not given
Token Factory: per-token pricing from $0.010/M tokens (BGE-M3) to $0.700/M tokens (GLM 5.2) for input, with output tokens at higher rates; image generation from $0.0005 per image (FLUX.1); speech-to-text from $0.0002/min; text-to-speech at $0.99/M characters; Dedicated GPU: H100 from $2.10/hr, H200
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
Token Factory: serverless inference with 30+ open models through single OpenAI-compatible key
AgentOS: governed agent execution with tools, approvals and audit trails
OpenAI-compatible API for text, vision, image, video and audio models
Dedicated Endpoints with reserved GPU capacity and autoscaling
AI Factory for VPC, on-premise and air-gapped private cloud deployment
Fine-tuning with managed runs supporting SFT, DPO and GRPO
Multi-model inference with intelligent routing and model switching
Up to 99.9% uptime SLA with GDPR and SOC 2 Type II compliance
Support for multiple runtimes and hardware accelerators including NVIDIA and AMD
Blueprints library with ready-to-run templates for inference and fine-tuning
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
Single OpenAI-compatible API key across 30+ models with no ecosystem lock-in
Transparent pricing at credible market rates with automatic repricing as market moves
Unified path from serverless inference to dedicated capacity to private cloud on same platform
Multi-architecture fleet using NVIDIA and AMD for high utilization and lower cost per token
Agent-native platform with governed execution, tool calls and audit trails for production workloads
FlexAI