fal

fal (Features and Labels, Inc.) is a generative media platform providing the world's fastest API for running image, video, audio, and multimodal generative AI models. Through a unified queue-based REST API at https://queue.fal.run, plus realtime WebSocket and SSE streaming surfaces, fal serves 1,000+ production models — including FLUX, Veo 3, Kling, Wan, Seedream, Nano Banana, and Stable Diffusion — on autoscaling GPU infrastructure. fal Serverless lets developers ship custom models with `@fal.function` / `fal.App` / BYO containers, while fal Compute provides dedicated H100/H200/A100/B200 instances. Trusted by Canva, Perplexity, Poe, and 1.5M+ developers; Series D funded ($140M, Sequoia-led, December 2025); SOC 2 with 99.99% uptime.

9 APIs 18 Features

AIArtificial IntelligenceGenerative AIGenerative MediaImage GenerationVideo GenerationAudio GenerationInferenceServerlessGPUMCP

APIs

fal Model APIs

Unified queue-based REST API for invoking 1,000+ generative image, video, audio, and multimodal models hosted on fal's inference infrastructure. Submit a request to `https://que...

fal Realtime API

WebSocket-based realtime inference for ultra-low latency interactive generative experiences such as LCM/SDXL sketch-to-image, live-portrait, and realtime upscaling. Bi-direction...

fal Streaming API

HTTP streaming endpoint (`/{model-id}/stream`) that emits progressive partial outputs as a model runs — used for LLM/VLM token streams, incremental video frames, and step-by-ste...

fal Storage API

REST endpoints for uploading binary inputs (images, audio clips, reference frames, control maps) to fal's CDN so they can be referenced by URL when invoking model APIs. Issues s...

fal Serverless Platform API

Programmatic management of custom fal Serverless applications — list, inspect, deploy, scale, and monitor user-defined GPU functions deployed with `@fal.function`, `fal.App`, or...

fal Models Catalog API

Read-only discovery endpoints for browsing fal's 1,000+ production model catalog, including model metadata, capability tags, pricing per output, supported parameters, example in...

fal Compute API

Provision and manage dedicated GPU instances (H100, H200, A100, B200) with full SSH access for training, fine-tuning, and persistent workloads. Hourly or per-second billing with...

fal API Keys API

Manage fal API keys — create, list, scope, and revoke keys used to authenticate against the Model, Storage, Serverless, and Compute APIs via the Authorization: Key $FAL_KEY header.

fal Usage and Billing API

Programmatic access to usage metrics, per-model spend, GPU-second consumption, and invoicing history. Surfaces the same data shown on the fal dashboard so platform teams can pip...

Collections

fal Model APIs

POSTMAN

fal Serverless Platform API

POSTMAN

fal Storage API

POSTMAN

fal Model APIs

OPEN

fal Serverless Platform API

OPEN

fal Storage API

OPEN

Arazzo Workflows

fal Upload, Run Image-To-Image With Webhook

Upload a reference image, submit an image-to-image job with a webhook, and confirm queue acceptance.

ARAZZO

fal Queue Inference

Submit a model inference job, poll the queue until it completes, then fetch the result.

ARAZZO

fal Serverless App Discovery

List deployed Serverless apps, then fetch full metadata and scaling for the first one.

ARAZZO

fal Serverless App Files Inspection

Confirm a Serverless app exists, then list files on its persistent /data volume.

ARAZZO

fal Set And Verify Serverless Secret

Create or replace a Serverless secret, then list secret names to confirm it is present.

ARAZZO

fal Streaming Inference

Run a model synchronously over the streaming endpoint to receive progressive output.

ARAZZO

fal Submit And Conditionally Cancel

Submit an inference job, check its status once, and cancel it if it has not finished.

ARAZZO

fal Upload Asset Then Run Inference

Upload a binary reference asset to the fal CDN, then run an image-to-X model against it.

ARAZZO

fal Webhook-Backed Submission

Submit an inference job with a webhook callback and confirm it was accepted into the queue.

ARAZZO

GraphQL

fal GraphQL API

fal is a fast serverless inference platform for AI models including image generation (Stable Diffusion, FLUX, Kling), video generation, speech, and custom models. The API covers...

GRAPHQL

Pricing Plans

Fal Ai Plans Pricing

2 plans

PLANS

Rate Limits

Fal Ai Rate Limits

5 limits

RATE LIMITS

FinOps

Fal Ai Finops

FINOPS

Features

Unified queue-based REST API at https://queue.fal.run/{model-id} for 1,000+ generative models

Image generation models — FLUX (Schnell, Dev, Pro, Kontext Pro), Seedream V4, Nano Banana, Qwen, SDXL, SD3, Ideogram, Recraft

Video generation models — Veo 3, Kling 2.5 Turbo Pro, Wan 2.5, Seedance 2.0, Ovi, Hunyuan, Sora-class

Audio and voice models — Inworld TTS-1.5, ElevenLabs, MMAudio, MusicGen, Stable Audio

3D and multimodal models — TripoSR, Hunyuan3D, LivePortrait, FaceChain

Synchronous, asynchronous queue, server-sent streaming, and WebSocket realtime invocation modes

Webhook callbacks for queue completion with HMAC signature verification

File uploads / CDN storage at https://v3.fal.media with signed upload URLs

fal Serverless — `@fal.function`, `fal.App`, BYO container deployment with autoscaling from 0 to thousands of GPUs

fal Compute — dedicated H100/H200/A100/B200 instances with SSH and per-second billing

Per-output billing (image, video second, audio minute) plus per-second GPU billing for custom deployments

99.99% uptime SLA, SOC 2 compliance, private endpoints, and enterprise support

Proprietary Inference Engine — up to 10x faster than reference implementations

Official SDKs for Python (fal-client), JavaScript/TypeScript (@fal-ai/client), Swift, Java/Kotlin, Dart

fal CLI for serverless deploy / run / apps / secrets / auth

fal MCP Server exposing all 1,000+ models to AI assistants via the Model Context Protocol

ComfyUI and Blender extensions, plus Terraform provider for infra-as-code

Day-zero launch partner for major model releases (FLUX, Veo, Kling, Seedance, Wan, etc.)

GitHubOrganization

Sources

opencollection: 1.0.0
info:
  name: fal Storage API
  version: v1
request:
  auth:
    type: apikey
    key: Authorization
    value: '{{Authorization}}'
    placement: header
items:
- info:
    name: Storage
    type: folder
  items:
  - info:
      name: Initiate Asset Upload
      type: http
    http:
      method: POST
      url: https://rest.alpha.fal.ai/storage/upload/initiate
      body:
        type: json
        data: '{}'
    docs: 'Request a short-lived signed URL to which the client should PUT the file bytes. Returns both the upload URL and
      the public CDN URL that should be passed to model APIs once the upload completes.

      '
bundled: true