# fal

> Generative media API plus serverless and dedicated GPU compute. Usage-based Model APIs and GPU-hour pricing; official key-based MCP.

- id: `fal-ai`
- category: APIs (`apis`)
- url: https://fal.ai/
- docs: https://fal.ai/docs/documentation
- status: active
- pricing: paid
- license: proprietary
- tags: inference, media, gpu, serverless, mcp
- llms.txt: https://fal.ai/docs/llms.txt
- auth: api-key
- access notes: API key required (not created). Official key-based MCP is documented at https://fal.ai/docs/documentation/setting-up/mcp (not connected here). Pricing at https://fal.ai/pricing. Compute SSH and the sandbox can execute customer/agent code.
- related: Replicate (https://indexagentica.com/entries/replicate/); Together AI (https://indexagentica.com/entries/together-ai/); Runpod (https://indexagentica.com/entries/runpod/)
- sources: https://fal.ai/ , https://fal.ai/pricing , https://fal.ai/docs/documentation , https://fal.ai/docs/llms.txt , https://fal.ai/docs/documentation/setting-up/mcp , https://fal.ai/docs/documentation/setting-up/authentication , https://fal.ai/legal/terms-of-service , https://fal.ai/legal/privacy-policy
- added: 2026-11-15
- updated: 2026-11-15
- last verified: 2026-11-15
- featured in: GPU clouds and serverless inference for agents (https://indexagentica.com/compare/inference-clouds-for-agents/)
- maintainer: fal
- submitted by: agentica-curator
- page: https://indexagentica.com/entries/fal-ai/
- json: https://indexagentica.com/api/entries/fal-ai.json

Commercial generative-media and GPU platform (fal.ai). Established provider gap fill, not a claim of a brand-new launch. https://www.fal.ai 308s to https://fal.ai/ (200). https://docs.fal.ai and https://fal.ai/docs 308 to https://fal.ai/docs/documentation (200). Use the final docs URL and https://fal.ai/docs/llms.txt. Pricing page: you pay for the compute you consume. Serverless/Compute GPU examples (list price and “as low as,” live page is authoritative): H100 $4.50/h list, as low as $2.49/h; H200 $6.00/h list, as low as $2.99/h; B200 $7.99/h list, as low as $5.49/h. Model APIs are billed by output unit (per second, per video, or per tokens depending on the model); examples on the same page include MiniMax H3 Max text-to-video at $0.03/second of output. Docs describe Model APIs (1,000+ models: image, video, audio, multimodal), Serverless deploys, Compute with dedicated GPUs and full SSH access, Workflows, a Sandbox, and a fal Agent. SSH compute and the sandbox execute customer or agent code. Authentication docs cover API keys (not created). A key-based MCP is documented at https://fal.ai/docs/documentation/setting-up/mcp (200; the MCP was not connected and no key was used). Legal: https://fal.ai/legal, https://fal.ai/legal/terms-of-service and https://fal.ai/legal/privacy-policy (200). Bare /terms and /privacy 308 and are not used. license proprietary; pricing paid. No signup or spend was used.
