# Run raw prompts (/docs/ai-visibility/prompt-runs)



Use `POST /v1/prompts/run` when you need answers and cited sources but do not need mention detection, position, competitor matching, or sentiment.

```bash
curl -X POST https://api.trackee.dev/v1/prompts/run \
  -H "x-access-key: YOUR_ACCESS_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "How should a startup choose project management software?",
    "engines": ["chatgpt", "gemini"],
    "web_search": true
  }'
```

Raw prompt runs cost 3 credits per engine. They do not create visibility snapshots, even when the prompt matches one saved on a brand.

Add the optional `models` object to choose a supported model per engine. Use [`GET /v1/models`](/docs/api/models/listModels) to discover the current choices.

Choose a [live visibility check](/docs/ai-visibility/live-checks) instead when you need to measure a specific brand or store the result in its history.

See the complete [`POST /v1/prompts/run` reference](/docs/api/prompts/runPrompt).