# fal

**Canonical:** https://apis.io/providers/fal-ai/  
**Website:** https://fal.ai  
**APIs profiled:** 12

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.

## Kin Score — 69.3 / 100 (exemplar)

Scored 2026-08-20 under rubric 0.12.0. Trend: flat (+0.0 from 69.3).

| Facet | Score |
|---|---|
| Discoverability | 81.5 |
| Contract Quality | 77.0 |
| Governance | 28.0 |
| Contract Governance | 28.0 |
| Operational Transparency | 50.0 |
| Developer Ergonomics | 78.6 |
| Commercial Clarity | 81.6 |
| Access Clarity | 81.6 |

## Agent readiness — 49.1 (agent-ready)

| Dimension | Value |
|---|---|
| Spec Presence | yes |
| Agentic Access | derived |
| Reversibility Documented | documented |
| MCP Server | verified |
| Auth Clarity | yes |
| Idempotency | no |
| Error Semantics | documented |
| OpenAPI Examples | partial |
| Rate Limit Signal | documented |
| Event Surface Described | derived |
| Agent Skills | no |
| Well Known Catalog | no |
| Consent Identity | no |
| Agent Card | no |
| Dry Run Mode | no |

## Access

Paid · Self-serve signup — onboarding: self-serve, pricing: paid, trial: no (confidence: high).

## APIs (12)

- **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 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...
- **fal Apps API** — List and inspect deployed Serverless apps.
- **fal Files API** — Manage files on persistent Serverless `/data` volumes.
- **fal Queue API** — Submit, inspect, and cancel model inference jobs.
- **fal Secrets API** — Manage per-org secrets injected into Serverless runs.
- **fal Storage API** — Upload binary assets to the fal CDN.
- **fal Streaming API** — Server-sent streaming of incremental model output.

## MCP servers (1)

- **fal MCP Server**

## Agentic access (1)

- **Fal Ai Agentic Access** — 11 operations · 5 acting

## Security (3)

- **Fal Ai Authentication** — apiKey · 1 scheme
- **Fal Ai Domain Security** — TLSv1.3 · HSTS · DNSSEC · DMARC
- **Fal Ai Trust Center** — SOC 2 Type II

## Plans (1)

- **Fal Ai Plans Pricing**

## Tags

Artificial Intelligence, Generative AI, Generative Media, Image-Generation, Video Generation, Audio Generation, Inference, Serverless, GPU, MCP

---

Profiled by [API Evangelist](https://apievangelist.com) and published on [APIs.io](https://apis.io/providers/fal-ai/). Scores are computed from the provider's own public artifacts under a published rubric.