Skip to content

What are Predictions

Understanding predictions (tasks) on Get3W.

Overview

A prediction is a single request to an AI model to generate content. When you submit a task through the API, web interface, or any SDK, Get3W creates a prediction that tracks the generation process.

Prediction Lifecycle

Created → Processing → Completed / Failed
StatusDescription
createdRequest received, task queued
processingAI model is actively generating content
completedGeneration successful, outputs available
failedAn error occurred during generation

Prediction Object

json
{
  "id": "pred_abc123",
  "model": "get3w/flux-dev",
  "status": "completed",
  "input": {
    "prompt": "A cat wearing a space suit"
  },
  "outputs": [
    "https://cdn.get3w.com/outputs/image-xxxxx.png"
  ],
  "timings": {
    "inference": 2500
  },
  "created_at": "2024-01-01T12:00:00.000Z",
  "completed_at": "2024-01-01T12:00:03.000Z"
}

Key Fields

FieldDescription
idUnique identifier for the prediction
modelThe model that was used
statusCurrent status of the prediction
inputThe parameters you submitted
outputsArray of URLs to generated content
timingsPerformance metrics (inference time in ms)
created_atWhen the prediction was created
completed_atWhen the prediction finished
errorError message (only present if failed)

Output Retention

Generated content is available for 7 days after creation. Download any files you want to keep permanently.

Retrieving Predictions

By ID

bash
curl "https://api.get3w.com/api/v3/predictions/pred_abc123" \
  -H "Authorization: Bearer YOUR_API_KEY"

Via Webhook

Set a webhook_url when submitting a task to receive a POST notification when the prediction completes:

json
{
  "prompt": "A cat in space",
  "webhook_url": "https://your-server.com/webhook"
}

Next Steps

Released under the MIT License.