Groq
Hosted language and speech models through GroqCloud, with familiar API interfaces.
- OpenAI-compatible interfaces for supported endpoints
- Language models and speech transcription
- Model-specific prices and request limits
GroqCloud provides hosted inference on Groq's infrastructure. Applications call supported language or speech models through the service rather than managing model servers.
Building with GroqCloud
The API supports common OpenAI client libraries with a different endpoint and API key. The model catalogue distinguishes production models from preview models and lists the limits for each. Check the chosen model's available tools and modalities before moving an existing integration.
How pricing works
Language-model rates depend on the model and separate input from output tokens. Speech endpoints use audio-based pricing. Some catalogue entries require a sales agreement instead of a public self-service rate. Free access and developer plans have different capacity limits.