Developer Quickstart

Authentication All API calls require a FAL_KEY header. Generate your key from the fal.ai dashboard.

# Install the fal client
pip install fal-client

import fal_client

result = fal_client.run(
    "fal-ai/flux/dev",
    arguments={"prompt": "A futuristic cityscape at sunset"}
)
print(result["images"][0]["url"])

Key Concepts

Model ID: Every model has a unique slug (e.g., fal-ai/kling-video/v1.6/pro/image-to-video). Browse all IDs in the Model Gallery.
Serverless Inference: Stateless, pay-per-output calls — no GPU provisioning needed.
Compute Clusters: For large-scale training or fine-tuning, spin up dedicated GPU clusters (H100/H200/B200) via the Compute API.
LoRA & Fine-tuning: Upload custom weights or trigger training jobs via the Training API endpoint.

SDKs & Integrations

Official SDKs: Python, JavaScript/TypeScript
REST API compatible with any HTTP client
Workflow builder and Sandbox available in the dashboard for no-code prototyping

Pricing Model

Serverless: Per-output pricing (pay only for what you generate)
Compute: Hourly GPU pricing starting at $1.89/hr for H100s
Enterprise: Reserved capacity with usage-based or flat-rate options