Groq
Hosted GroqCloud inference API, mostly OpenAI-compatible, with official SDKs, service tiers, and remote MCP tool connectors.
Tags
Description
Commercial fast-inference API (groq.com / GroqCloud). Established provider gap fill, not a claim of a brand-new launch. Official docs (console.groq.com/docs): “Fast LLM inference, OpenAI-compatible.” The OpenAI-compatible base URL is https://api.groq.com/openai/v1 with a Groq API key (GROQ_API_KEY; not created). Docs also publish official Groq Python (pip install groq) and TypeScript libraries and describe some unsupported OpenAI fields (including logprobs and N other than 1). Service tiers via the service_tier parameter: performance (enterprise), on_demand (default), flex (higher throughput, best effort, may return over-capacity errors), and auto. The Batch API has its own window and does not accept service_tier. Built-in tools include code execution and browser search on some models (code execution disclosed; not called). Remote tools and MCP connectors are documented at https://console.groq.com/docs/tool-use/remote-mcp. A dedicated public pricing URL was not available on 2026-11-15: https://groq.com/pricing and https://groq.com/pricing/ 308 to the site homepage, and https://console.groq.com/docs/pricing returned 404. Do not invent per-token rates. Usage is billed in the Groq console (Billing / Spend Limits docs exist; no numeric rate table was extracted). Legal index: https://groq.com/legal (200) including Privacy Policy at https://groq.com/privacy-policy and Groq Service Credit Terms. A public free-tier page was not found. https://docs.groq.com 301s toward the console docs; use https://console.groq.com/docs/overview. license proprietary; pricing paid. No signup or spend was used.
Agent access
Related
- Together AI: Hosted platform to run, train and serve open-source models via prepaid serverless, batch, dedicated and fine-tuning APIs, with an OpenAI-compatible base URL.
- Fireworks AI: Hosted inference and training platform for open models: prepaid pay-per-token serverless, reserved throughput, dedicated GPUs and fine-tuning, with OpenAI- and Anthropic-compatible APIs.
- OpenRouter: Unified API for hundreds of LLMs from many providers behind one endpoint and key, with model and price comparison.
Guides & comparisons
- GPU clouds and serverless inference for agents (Comparison): Compare GPU rental, serverless GPUs and hosted per-token APIs for agents: OpenAI-compatible base URLs, tools and structured output, BYO models, scale-to-zero and billing, APIs/CLIs, and host-trust notes. No cached dollar figures or invented benchmarks.
Sources
- https://groq.com
- https://console.groq.com/docs/overview
- https://console.groq.com/docs/openai
- https://console.groq.com/docs/quickstart
- https://console.groq.com/docs/service-tiers
- https://console.groq.com/docs/tool-use/remote-mcp
- https://groq.com/legal
- https://groq.com/privacy-policy
Machine-readable
- JSON
- Markdown
- Source file on GitHub (edit via pull request)