Skip to content

askDoc.md

Latest commit

 

History

History
66 lines (56 loc) · 8.78 KB

askDoc.md

File metadata and controls

66 lines (56 loc) · 8.78 KB

Descriptions of the ask method

🚀 Sends a request to the Grok API and receives a response. There is an asynchronous variant async_ask.

📨 Takes:

  • 📜 message: The text of the request for the model.
  • ⚙ **kwargs: Additional parameters for configuration.

🎯 Returns:

Complete Parameter List:

Parameter Type Description Default
message str Required. The request text for Grok. -
history_id str Identifier in the history dictionary (if auto saving history is used) None
new_conversation bool Whether to use the new chat URL when sending a request to Grok (does not affect the built-in History class). False
timeout int Timeout for one wait to receive a response (seconds). 45
temporary bool Indicates if the request or session is temporary. False
modelName str Name of the AI model (e.g., "grok-3"). "grok-3"
images str_path / str_base64 / BytesIO / List[str] / List[BytesIO] Either the path to an image, a base64-encoded image, or a BytesIO (or a list of any of these) to send. fileAttachments must not be used. None
fileAttachments List[str] List of file attachments (keys: name, content). []
imageAttachments List[] List of image attachments (keys: name, content). []
customInstructions str Additional instructions for the model. ""
deepsearch_preset str Preset for deep search. ""
disableSearch bool Disable search in model responses. False
enableImageGeneration bool Enable image generation. True
enableImageStreaming bool Enable image streaming. True
enableSideBySide bool Enable side-by-side information display. True
imageGenerationCount int Number of images to generate. 4
isPreset bool Indicates if the message is preset. False
isReasoning bool Enable reasoning mode for the model. False
returnImageBytes bool Return images as bytes. False
returnRawGrokInXaiRequest bool Return raw model output. False
sendFinalMetadata bool Send final metadata with the request. True
forceConcise bool Force concise responses. True
disableTextFollowUps bool Disable text follow-ups. True
webpageUrls List[str] List of webpage URLs. []
disableArtifact bool Disable artifact flag. False
toolOverrides Dict[str, Any] Override tool settings. {}
responseModelId str Model ID for response metadata. "grok-3"

⚙️ toolOverrides:
Some settings may not be recognized by the server or may be automatically overridden by the system.
For fine-tuning behavior, you can use the toolOverrides parameter, passing a dictionary with overrides.
For example, the official website sometimes automatically applies the following toolOverrides block:

toolOverrides = {
  "imageGen": False,
  "webSearch": True,
  "xSearch": True,
  "xMediaSearch": True,
  "trendsSearch": True,
  "xPostAnalyze": True
}

These options allow enabling or disabling various tools and model functions.

💡 It is important to understand that these parameters are obtained by reverse engineering browser requests. And, perhaps, some of them may not yet have functionality, especially considering the freshness of the Grok3 model

❗ Descriptions of those parameters whose functionality could not be confirmed in testing are based on similar parameters in the official xAI API documentation.

🛠️ You can contribute by simply experimenting with different options!