DeepInfra

Serverless inference API for running open-source LLMs cheaply

Code & DevelopmentPaid

DeepInfra is a serverless inference platform that provides API access to hundreds of open-source machine learning models—including Llama, Mistral, Mixtral, Whisper, and Stable Diffusion—at a fraction of the cost of proprietary API providers. Models are available as REST APIs with OpenAI-compatible endpoints, enabling drop-in replacement for applications built on the OpenAI SDK. DeepInfra's auto-scaling infrastructure handles bursts without pre-provisioning, with pay-per-token pricing and no minimum commitments.

Key Features

• ✓100+ open-source models
• ✓OpenAI-compatible API
• ✓Serverless scaling
• ✓Pay-per-token
• ✓Whisper/Stable Diffusion
• ✓Low cost