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

# Response options — Markdown, raw responses and include flags

> Which payload.include flags each engine honors, which engines return Markdown without asking, and what happens to a flag an engine does not support.

`payload.include` adds optional fields to a task's response. Each flag is a boolean, and each engine honors only some of them.

```json theme={"theme":"css-variables"}
{
  "taskType": "PERPLEXITY",
  "payload": {
    "prompt": "best CRM for a small team",
    "country": "US",
    "include": { "markdown": true, "rawResponse": true }
  }
}
```

## Flags

| Flag | Adds |
| - | - |
| `markdown` | The answer as Markdown, in `markdown` (`aioverview.markdown` on `GOOGLE` and `BING_SEARCH`). |
| `rawResponse` | The engine's raw output, as `rawResponse` or `rawContent` (see below). |
| `searchQueries` | The searches the engine ran, in `searchQueries`. |
| `ads`, `shopping` | Ads and product cards. `CHATGPT` only. |
| `html` | The screen the engine showed. See [HTML pages](/engines/html-pages.md). |
| `analysis` | Brands, recommendations, sentiment and source types. See [Answer analysis](/engines/answer-analysis.md). |
| `entityDetails` | The detail panel of each card in the answer, free. `CHATGPT` only. See [ChatGPT](/engines/chatgpt.md#card-details). |
| `citationAnalysis` | A Citation Analysis of the answer. See below. |

## Support matrix

✓ honored, **always** returned without the flag, — no effect (ignored), ✗ rejected.

| taskType | `markdown` | `rawResponse` | `searchQueries` | `html` | `analysis` | `citationAnalysis` |
| - | - | - | - | - | - | - |
| `CHATGPT` | ✓ (default on) | ✓ | ✓ (default on) | ✓ | ✓ | ✓ |
| `GEMINI` | ✓ | ✓ | — | ✓ | ✓ | ✓ |
| `PERPLEXITY` | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| `GOOGLE` | ✓ | ✓ `rawContent` | — | ✓ | ✓ | ✓ |
| `AIMODE` | **always** | ✓ `rawContent` | — | ✓ | ✓ | ✓ |
| `BING_COPILOT` | ✓ | ✓ | always | ✗ | ✗ | ✓ |
| `NAVER_AI_BRIEF` | **always** | ✓ `rawContent` | — | ✗ | ✓ | ✓ |
| `NAVER_AI_TAB` | **always** | ✓ `rawContent` | — | ✗ | ✓ | ✓ |
| `META_AI` | **always** | ✓ | always | ✓ | ✗ | ✗ |
| `QWEN` | **always** | ✓ | always | ✗ | ✗ | ✗ |
| `DOUBAO` | **always** | ✓ | always | ✗ | ✗ | ✗ |
| `GOOGLE_SERP` | — | ✓ `rawContent` | — | ✗ | ✗ | ✗ |
| `BING_SEARCH` | ✓ | ✓ `rawContent` | — | ✗ | ✗ | ✓ |
| `NAVER_SERP`, `COMMERCE_SERP`, `COMMERCE_PRODUCT`, `REDDIT`, `WEB_SCRAPE`, `WEB_MAP`, `WEB_CRAWL`, `KEYWORD_VOLUME` | — | — | — | ✗ | ✗ | ✗ |

`ads` and `shopping` work on `CHATGPT` only, where they default to on; every other engine ignores them. `searchQueries` marked "always" means the engine returns the field whether or not you set the flag.

### Markdown without a flag

`AIMODE`, `NAVER_AI_BRIEF`, `NAVER_AI_TAB`, `META_AI`, `QWEN` and `DOUBAO` return `markdown` on every answer. For these engines Markdown is the answer's real shape (headings, lists, tables and cards in place), not a copy of `text`. `CHATGPT` also returns it by default; send `"markdown": false` to drop it.

`GOOGLE` also accepts the older flag `include.aioverview.markdown`, which does the same as `include.markdown`.

### Raw responses

`rawResponse` is off by default on every engine, including `CHATGPT`. Raw output arrives in one of two fields:

* `rawResponse`: an array of parsed stream events (`CHATGPT`, `GEMINI`, `PERPLEXITY`, `BING_COPILOT`, `META_AI`, `QWEN`, `DOUBAO`).
* `rawContent`: a string. For `GOOGLE`, `AIMODE`, `GOOGLE_SERP` and `BING_SEARCH` it is the full rendered results page HTML; for `NAVER_AI_BRIEF` and `NAVER_AI_TAB` it is Naver's original event stream.

These fields can run to hundreds of kilobytes, so request them only when you need them.

## Unsupported flags

What happens depends on the flag:

| Flag | On an engine that does not support it |
| - | - |
| `markdown`, `rawResponse`, `searchQueries`, `ads`, `shopping` | Silently ignored. No error and no field. |
| `html` | `422 EVIDENCE_UNSUPPORTED`. |
| `analysis` | `422 VALIDATION_ERROR` on `payload.include.analysis`. |
| `citationAnalysis` | `400 VALIDATION_ERROR` on `payload.include.citationAnalysis`. |
| `entityDetails` | `422 VALIDATION_ERROR` on `payload.include.entityDetails`. |
| `screenshot` | `400 VALIDATION_ERROR` on every engine. It is no longer available; use `html`. |

`html`, `screenshot` and `citationAnalysis` must be booleans (`400 VALIDATION_ERROR` otherwise), and so must `analysis` (`422 VALIDATION_ERROR`). All rejections happen at submit, before any credits are held.

## Citation Analysis

`include.citationAnalysis: true` starts a [Citation Analysis](/research/citation-analysis.md) of the answer once it completes. It is a separate, metered task, the answer's webhook waits for it, and reads embed it under `response.citationAnalysis`. It works on `CHATGPT`, `GEMINI`, `PERPLEXITY`, `GOOGLE`, `AIMODE`, `BING_SEARCH`, `BING_COPILOT`, `NAVER_AI_BRIEF` and `NAVER_AI_TAB`. The API key must also be allowed to submit `CITATION_ANALYSIS` (`403` otherwise). See [Citation Analysis](/research/citation-analysis.md) for the result and its price.

## Related

* [Engine overview](/engines/overview.md)
* [Answer analysis](/engines/answer-analysis.md)
* [HTML pages](/engines/html-pages.md)


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