N8N Integration
n8n community node for WaveSpeedAI. Generate images and videos, or run any model on the WaveSpeed catalog, from an n8n workflow without writing HTTP requests.
Installation
- In n8n, open Settings -> Community Nodes
- Click Install and enter
@wavespeed/n8n-nodes-wavespeed - Search for “WaveSpeed” in the node picker
Requires n8n 1.x and Node.js 20 or newer.
Setup
- Get your API key
- In n8n, add credentials of type WaveSpeed API
- Paste your API key and save
The credential test calls GET /api/v3/balance, so a green check also confirms the key has API access.
Operations
The WaveSpeed node submits a prediction and waits for the result. Each item’s output contains urls, outputs, id, model and timings.
| Operation | Purpose |
|---|---|
| Generate Image | Image from a text prompt. Default model bytedance/seedream-v5.0-pro; optional resolution, aspect ratio and extra inputs as JSON |
| Generate Video | Video from a text prompt. Optional duration and extra inputs as JSON |
| Run Model | Run any model on wavespeed.ai/models with its model ID and the full request body as JSON — image-to-image, image-to-video, LoRA, upscaling, lipsync and more |
Shared options:
- Download Output — attach the generated files as n8n binary data (
data,data_1, …) for the next node - Poll Interval (Seconds) — how often to check progress
- Timeout (Seconds) —
0waits until the task finishes; a positive value fails the node after that long while the prediction keeps running server-side
Transient poll failures (network errors, HTTP 429 and 5xx) are retried automatically.
Warning: Leave n8n’s Settings -> Retry On Fail switched off for generation operations. n8n retries by re-running the whole node, which submits a new, separately billed prediction. Use Continue On Fail if you need the task ID of a failed item.
Example Workflows
Prompt to image:
Manual Trigger -> WaveSpeed (Generate Image) -> Use ResultRun any model with your own inputs:
Set (build JSON body) -> WaveSpeed (Run Model) -> Use ResultTroubleshooting
| Problem | Solution |
|---|---|
| Invalid API key | Recreate the key and update credentials |
| Parameter errors | Check the model’s input schema on its model page and pass the fields in Additional Inputs or the Run Model body |
| Node times out | Set Timeout to 0 or a longer value; the prediction keeps running and the error names the task ID |
| Charged twice for one item | Turn off Retry On Fail (see above) |