> ## Documentation Index
> Fetch the complete documentation index at: https://docs.markifact.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Audit Llm Answer

> Audit how ChatGPT, Claude, Gemini or Perplexity answer a prompt about a brand, product or category, for AI visibility, answer engine optimization (AEO) and generative engine optimization (GEO): ret...

Audit how ChatGPT, Claude, Gemini or Perplexity answer a prompt about a brand, product or category, for AI visibility, answer engine optimization (AEO) and generative engine optimization (GEO): returns the answer, the sources it cited and the brands it names. Only for measuring AI answers, never as a general assistant to draft content or answer the user: each call is billed to the user's DataForSEO balance plus the model provider's token cost.

| | |
| - | - |
| **App** | DataForSEO |
| **Operation ID** | `dataforseo_audit_llm_answer` |
| **Type** | Action |
| **Connection** | `dataforseo` (required) |
| **Credits per run** | Free |
| **Agent / MCP tool** | Yes |

## Inputs

| Field | Type | Required | Default | Description |
| - | - | - | - | - |
| `provider` | string or SelectableOption | No | `chat_gpt` | AI provider to ask: 'chat\_gpt' (OpenAI, default), 'claude' (Anthropic), 'gemini' (Google) or 'perplexity'. |
| `model` | string | No | - | Model name as listed by dataforseo\_list\_llm\_models, e.g. gpt-4.1-mini, gpt-5.5, claude-sonnet-5, gemini-2.5-flash, sonar-pro. Leave empty for the provider default (gpt-4.1-mini, claude-haiku-4-5, gemini-2.5-flash, sonar). Reasoning models such as the gpt-5 family cost more and cannot be forced to search the web. |
| `prompt` | string | Yes | - | The question or task to send, e.g. 'What are the best project management tools for agencies?' |
| `system_message` | string | No | - | Optional instructions that set the role, tone or format of the answer. |
| `web_search` | string or SelectableOption | No | `auto` | 'auto' (default) lets the model decide whether to search the web, 'always' forces a web search so the answer cites live sources (chat\_gpt non-reasoning models such as gpt-4.1-mini, and claude), 'off' answers from training data only. Perplexity always searches the web. |
| `country` | string or SelectableOption | No | - | Two-letter country code (US, GB, DE...) used to localise the web search; a DataForSEO country location code such as 2840 also works. Not supported by gemini. Leave empty for the provider default. |
| `max_output_tokens` | integer | No | - | Maximum length of the answer in tokens (16 to 4096). Leave empty for the model default. Reasoning models such as the gpt-5 family and the o-series do not accept it. |
| `temperature` | number | No | - | Randomness of the answer from 0 (deterministic) to 2 (creative). Leave empty for the provider default. |

### SelectableOption

| Field | Type | Required | Default | Description |
| - | - | - | - | - |
| `value` | string | Yes | - | |
| `label` | string | Yes | - | |

## Output

**Type**: `List[Dict]`

Returns a single row: provider, model\_name, answer (markdown), sources (list of title, domain, url), cited\_domains (comma-separated unique domains), fan\_out\_queries, web\_search (whether the model searched), input\_tokens, output\_tokens, reasoning\_tokens, money\_spent (provider cost in USD), datetime.

**Fields**: `provider`, `model_name`, `answer`, `sources`, `cited_domains`, `web_search`, `money_spent`

**Example**:

```json theme={"dark"}
[
  {
    "provider": "chat_gpt",
    "model_name": "gpt-4.1-mini",
    "answer": "The most popular backlink checkers are Ahrefs and Semrush.",
    "cited_domains": "ahrefs.com, semrush.com",
    "web_search": true
  }
]
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.