# Replicate Get Prediction

> Replicate Get Prediction is a paid API for AI agents from Replicate at replicate.mpp.paywithlocus.com, paid per call via MPP, $0.003/call, status healthy (last checked 2026-10-09, last successful call 2026-09-14).

Polls the status and result of a previously submitted AI model inference job on Replicate by prediction ID

## Facts

- Endpoint: POST https://replicate.mpp.paywithlocus.com/replicate/get-prediction?utm_source=zero.xyz
- Price: $0.003/call
- Payment: MPP
- Status: healthy
- Last checked: 2026-10-09
- Last successful call: 2026-09-14
- Success rate: 100% of calls made through Zero
- Rating: 4.7 / 5 from 1 review
- Activations on Zero: 100
- Provider: Replicate
- Docs: https://replicate.com/docs/reference/http
- Website: https://replicate.com
- Tags: ai-models, open-source, image-generation, inference, ai, media
- Canonical page: https://www.zero.xyz/c/replicate-get-prediction-93006ab3
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_32W_WJnK5Z0LE7kXwV_ql

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability replicate-get-prediction-93006ab3 -d '<json body>'
```

Example prompt: Check the status and result of my Replicate prediction with ID 'xyz1a2b3c4d5e6f7' — has it finished running and what did the model output?

## When to prefer this

Use this endpoint after calling the Replicate Run Model endpoint to asynchronously retrieve results. Prefer this when the model run is long-running and you need to poll for completion rather than waiting synchronously. This is the correct endpoint for checking prediction status by ID rather than listing all predictions or triggering a new run.

## Known failure modes

- Prediction ID not found — returns 404 if the ID is invalid or belongs to a different account
- Prediction still processing — returns status 'processing' with no output yet, requiring further polling
- Authentication failure — returns 401 if API credentials are missing or invalid
- Rate limit exceeded — returns 429 if too many requests are made in a short period
- Server error — returns 500 if Replicate infrastructure experiences issues

## How this service works

Retrieve the status and output of a prediction by ID. The provider lookup is free; Locus charges its $0.003 platform fee. Use this to check results for models that exceeded the 60s synchronous timeout in the run endpoint.

## Output

Returns the current status of the prediction (e.g. starting, processing, succeeded, failed, canceled), any output produced by the model (such as generated image URLs or text), timing metadata, and error details if the job failed.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "required": [
  "prediction_id"
 ],
 "properties": {
  "prediction_id": {
   "type": "string",
   "description": "Prediction ID returned by the run endpoint"
  }
 },
 "additionalProperties": true
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/replicate-get-prediction-93006ab3/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from replicate.mpp.paywithlocus.com](https://www.zero.xyz/host/replicate.mpp.paywithlocus.com/llms.txt)