Built by Metorial, the integration platform for agentic AI.
Run synchronous inference on any Fal.ai model endpoint with arbitrary input parameters. This is a generic tool for calling any model that doesn't have a dedicated tool, including 3D generation, image editing, upscaling, and other specialized models. Pass model-specific parameters directly and receive model output with downloadable files when produced.
Copy a public file of up to 90 MiB to Fal.ai CDN storage from a URL. The uploaded file can then be referenced by its CDN URL in model inference requests. Useful for providing input images, audio, or video files to Fal.ai models.
Search and discover available model endpoints on Fal.ai. Supports listing all models, searching by text query or category, and looking up specific models by endpoint ID. Returns model metadata including name, description, category, and status.
Transcribe audio files to text using Fal.ai speech recognition models like Whisper (Wizper). Supports speaker diarization, language detection, and word/segment-level timestamps. Provide an audio URL and receive a full transcription with optional metadata.
Generate images from text prompts or transform existing images using Fal.ai models such as FLUX, Stable Diffusion, Ideogram, Recraft, and more. Supports text-to-image and image-to-image generation with configurable parameters like image size, guidance scale, LoRA adapters, and safety settings. Runs synchronously and provides downloadable generated images. Model parameters vary; use search_models with includeSchema=true to inspect the selected endpoint.
Generate speech audio from text using Fal.ai text-to-speech models. Model-specific options support language selection and reference-audio voice cloning when available. Provides a downloadable audio file. Inspect endpoint inputs with search_models using includeSchema=true.
Generate videos from text prompts, images, or other videos using Fal.ai models such as Kling, LTX, Veo, Sora, and more. Supports text-to-video, image-to-video, and video-to-video transformations. Runs synchronously and provides a downloadable video. Prefer submit_queue_request for slow models; inspect endpoint parameters with search_models using includeSchema=true.
Submit an asynchronous inference request to Fal.ai's queue system. This is the recommended approach for long-running model inference. Returns a request ID for polling status or retrieving results later. Optionally provide a webhook URL to receive results automatically upon completion.
Check the status of an asynchronous queue request on Fal.ai and optionally retrieve the result. Use after submitting a request with the Submit Queue Request tool. Use action="result" after completion to retrieve model output and downloadable files. action="cancel" requests cancellation; running inference may still finish.
Get the current unit price, billing unit, and currency for fal.ai model endpoints before running paid inference. Discover endpoint IDs with search_models. Final cost can depend on resolution, duration, and model-specific billing rules.
Identify the authenticated fal.ai account and retrieve its credit balance. Requires an ADMIN-scoped API key; ordinary model inference and discovery use API-scoped keys.