fal-ai
Use when generating images or video with AI models via fal.ai — FLUX, Kling, Veo 2, Mochi, Hunyuan. Also use when doing real-time inference, queue management, or integrating fal.ai with Vercel AI SDK.
| Model | Source |
|---|---|
| sonnet | pack: ai |
Full Reference
┏━ 🔧 fal-ai ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓ ┃ GPU inference for 600+ media generation models ┃ ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛
fal.ai
Section titled “fal.ai”Unified API for 600+ production-ready image, video, audio, and 3D models. Runs on H100/H200/A100 GPUs with a queue system, real-time WebSocket mode, and Vercel AI SDK provider. Package: @fal-ai/client v1.9.4.
Quick Reference
Section titled “Quick Reference”| Fact | Value |
|---|---|
| Install | npm install @fal-ai/client |
| Env var | FAL_KEY |
| Queue base URL | https://queue.fal.run/{model_id} |
| FLUX schnell | $0.003/megapixel |
| FLUX.2 dev | $0.012/megapixel |
| FLUX.2 pro | $0.03/first MP + $0.015/extra MP |
| Kling 2.6 Pro | $0.07/second (audio off) · $0.14/second (audio on) |
| Veo 3 | $0.40/second |
| Wan 2.5 video | $0.05/second |
| Docs | https://docs.fal.ai |
Reference Index
Section titled “Reference Index”| I want to… | File |
|---|---|
| Install the client, set up auth, configure TypeScript | reference/setup.md |
| Generate images with FLUX, SDXL, or Ideogram | reference/image-generation.md |
| Generate video with Kling, Mochi, Hunyuan, or Veo | reference/video-generation.md |
| Use WebSocket real-time inference for fast/interactive models | reference/realtime.md |
| Manually manage queue: submit, poll status, retrieve result | reference/queue-management.md |
| Integrate fal.ai with Vercel AI SDK via @ai-sdk/fal | reference/ai-sdk-provider.md |
| Browse top models by category with pricing | reference/models-and-pricing.md |
Usage: Read the reference file matching your task. Each file is self-contained with working code examples and inline gotchas.