GroqCloud

Offering overview

Speed is the pitch at GroqCloud: open-weight models are served on Groq's own Language Processing Units rather than GPUs, and exposed through an OpenAI-compatible API. The catalog covers Llama 3.1 8B and 3.3 70B, OpenAI's GPT-OSS 20B and 120B (including a safeguard variant), Qwen, MiniMax M2.7 for enterprise accounts, Kimi K2, Whisper Large V3 for speech-to-text and Orpheus text-to-speech, plus Llama Prompt Guard models for moderation. Published list prices are per million tokens: Llama 3.1 8B Instant at $0.05 input / $0.08 output, GPT-OSS 20B at $0.075 / $0.30, GPT-OSS 120B at $0.15 / $0.60 and Llama 3.3 70B at $0.59 / $0.79. A batch API halves standard rates and prompt caching lowers input costs further. Beyond the shared cloud, Groq sells GroqMetal (dedicated bare-metal LPU capacity), GroqCore (a tested inference stack) and GroqAssured (governance and audit controls) for enterprises. Infrastructure spans the United States, Canada, Finland, the United Kingdom, Saudi Arabia and Australia. Developers who need low per-token latency for chat, agents and real-time voice are the main audience, from free-tier experimenters to enterprise contracts.

Technical specifications

Containers

Unknown

Kubernetes

Unknown

Root access

Unknown

API access

Unknown

Autoscaling

Unknown

Persistent processes

Unknown

AI capabilities

Model serving

Llama 3.1 8B, Llama 3.3 70B, GPT-OSS 20B, GPT-OSS 120B, Qwen3, MiniMax M2.7, Kimi K2, Whisper Large V3, Orpheus, Llama Prompt Guard 2

AI frameworks

Not stated

Inference endpoints

Unknown

Fine-tuning

Unknown

Vector database

Unknown

Model catalog

Unknown

Monitoring

Not stated

Evaluation

Not stated

Security & compliance

Geography

Provider headquarters are separate from offering infrastructure and availability.

Available plans

Pricing can change. Promotional prices are shown separately from standard prices.