fal.ai

Overview

fal.ai provides instant serverless inference for AI models — no GPU provisioning, no containers, pay per inference. The API is simple: pick a model ID, pass inputs, get outputs. Supports async jobs with webhooks, real-time streaming, and file uploads. Best-in-class for Flux image generation, video generation, and audio models.

Install

npm install @fal-ai/client    # JavaScript/TypeScript
pip install fal-client         # Python

Authentication

export FAL_KEY="your-api-key"  # Get from fal.ai/dashboard