GroqCloud

What it does

GroqCloud is a hosted API that takes text, audio and image inputs and returns generated text, speech transcriptions, speech audio and moderation or safety classifications, depending on the model used. Developers call it with a Groq API key using the Groq Python or JavaScript libraries, or through OpenAI-compatible endpoints at api.groq.com. The model list includes large language models such as GPT-OSS 120B and 20B, Llama models, Whisper speech-to-text, Orpheus text-to-speech and Llama Prompt Guard classifiers.

Typical use is calling a chat completion from application code. The quickstart shows setting GROQ_API_KEY as an environment variable and sending a chat message to the llama-3.3-70b-versatile model in the retrieved example. The Vercel AI SDK, LiteLLM and LangChain are documented as third-party routes. The documentation also lists model discovery through the /openai/v1/models endpoint, which returns the active models and their IDs.

It suits developers building applications on open-weight models who need fast inference. Access requires creating an API key in the GroqCloud console. Some models are marked Enterprise or contact-sales, preview models should not be used in production, and deprecated models are unsupported or scheduled to lose support on model-specific dates. Pricing varies per model and unit, so readers should check the model table and billing documentation for current rates.

What it can do

• Chat completions with hosted open-weight language models
• Speech-to-text transcription with Whisper models
• Text-to-speech generation with Orpheus models
• Content moderation and prompt-injection classification with Prompt Guard and safeguard models
• Listing active models through the models API endpoint
• OpenAI-compatible API access

Works with

• Vercel AI SDK
• LiteLLM
• LangChain

Example workflows

Application chat completion

Illustrative example. Input: a user prompt in application code. Steps: install the Groq Python library, set GROQ_API_KEY, call chat.completions.create with a model ID such as llama-3.3-70b-versatile. Output: the model's text response.

Model discovery

Illustrative example. Input: a valid GROQ_API_KEY. Steps: send a GET request to https://api.groq.com/openai/v1/models with a Bearer token. Output: a JSON list of active model IDs to choose from in code.

Before you choose

• Preview models are for evaluation only and may be discontinued at short notice.
• Deprecated models are unsupported or scheduled to lose support; check the model-specific deprecation date before building on them.
• Pricing is listed per model and per unit (tokens, hours, characters), and some models require contacting sales.

Listed under

• Coding
• Sandboxes and infrastructure
• AI inference
• Cloud
• Developers
• Fast inference
• Inference
• Infrastructure

Get started

Create an API key in the GroqCloud console, then set GROQ_API_KEY and make a request.

1. Create an API key in the GroqCloud console.
2. Set the key as the GROQ_API_KEY environment variable.
3. Install the Groq Python library with pip and run a chat completion, or use the AI SDK provider.

Free provides recurring rate-limited access. Developer accepts credit cards, US bank accounts and SEPA debit accounts, and bills model usage at payment thresholds or the billing cycle. GPT-OSS 120B is listed at $0.15 per million input tokens and $0.60 per million output tokens. Audio and other models use their own units. Some models require Enterprise access.

Source: console.groq.com/docs/billing-faqs, checked 1 Nov 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 1 Nov 2026. Each source supports part of the overview; availability and prices can change.

• Billing FAQs - GroqDocs ↗console.groq.com
• Supported Models - GroqDocs ↗console.groq.com
• Quickstart - GroqDocs ↗console.groq.com

Company website

console.groq.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 1 Nov 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.