GenerationConfig

Configuration options for model generation and outputs. Not all parameters are configurable for every model.

Agentic AIArtificial IntelligenceCode GenerationEmbeddingsGenerative AIImage GenerationLLMMachine LearningMultimodal

Properties

Name	Type	Description
candidateCount	integer	Number of generated responses to return. Currently, this value can only be set to 1. If unset, this will default to 1.
stopSequences	array	The set of character sequences (up to 5) that will stop output generation. If specified, the API will stop at the first appearance of a stop_sequence. The stop sequence will not be included as part of
maxOutputTokens	integer	The maximum number of tokens to include in a response candidate. Note: The default value varies by model.
temperature	number	Controls the randomness of the output. Values can range from 0.0 to 2.0 inclusive. A higher value will produce responses that are more varied and creative, while a value closer to 0.0 will typically r
topP	number	The maximum cumulative probability of tokens to consider when sampling. The model uses combined Top-k and Top-p (nucleus) sampling. Tokens are sorted based on their assigned probabilities so that only
topK	integer	The maximum number of tokens to consider when sampling. Gemini models use Top-p (nucleus) sampling or a combination of Top-k and nucleus sampling.
responseMimeType	string	MIME type of the generated candidate text. Supported MIME types are text/plain (default), application/json (JSON response in the response candidates), and text/x.enum (for classification tasks).
responseSchema	object	Output schema of the generated candidate text. Schemas must be a subset of the OpenAPI schema and can be objects, primitives, or arrays. If set, a compatible responseMimeType must also be set. Compati
presencePenalty	number	Positive values penalize tokens that have already appeared in the generated text, increasing the likelihood of generating more diverse content.
frequencyPenalty	number	Positive values penalize tokens based on their frequency in the generated text, decreasing the likelihood of repeating the same content verbatim.
seed	integer	Seed used in decoding. If specified with the same seed and parameters, the model will attempt to return the same response.

View JSON Schema on GitHub

JSON Schema

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "#/components/schemas/GenerationConfig",
  "title": "GenerationConfig",
  "type": "object",
  "description": "Configuration options for model generation and outputs. Not all parameters are configurable for every model.",
  "properties": {
    "candidateCount": {
      "type": "integer",
      "format": "int32",
      "description": "Number of generated responses to return. Currently, this value can only be set to 1. If unset, this will default to 1."
    },
    "stopSequences": {
      "type": "array",
      "description": "The set of character sequences (up to 5) that will stop output generation. If specified, the API will stop at the first appearance of a stop_sequence. The stop sequence will not be included as part of the response.",
      "items": {
        "type": "string"
      }
    },
    "maxOutputTokens": {
      "type": "integer",
      "format": "int32",
      "description": "The maximum number of tokens to include in a response candidate. Note: The default value varies by model."
    },
    "temperature": {
      "type": "number",
      "format": "float",
      "description": "Controls the randomness of the output. Values can range from 0.0 to 2.0 inclusive. A higher value will produce responses that are more varied and creative, while a value closer to 0.0 will typically result in more straightforward responses from the model."
    },
    "topP": {
      "type": "number",
      "format": "float",
      "description": "The maximum cumulative probability of tokens to consider when sampling. The model uses combined Top-k and Top-p (nucleus) sampling. Tokens are sorted based on their assigned probabilities so that only the most likely tokens are considered."
    },
    "topK": {
      "type": "integer",
      "format": "int32",
      "description": "The maximum number of tokens to consider when sampling. Gemini models use Top-p (nucleus) sampling or a combination of Top-k and nucleus sampling."
    },
    "responseMimeType": {
      "type": "string",
      "description": "MIME type of the generated candidate text. Supported MIME types are text/plain (default), application/json (JSON response in the response candidates), and text/x.enum (for classification tasks)."
    },
    "responseSchema": {
      "type": "object",
      "description": "Output schema of the generated candidate text. Schemas must be a subset of the OpenAPI schema and can be objects, primitives, or arrays. If set, a compatible responseMimeType must also be set. Compatible MIME types: application/json.",
      "additionalProperties": true
    },
    "presencePenalty": {
      "type": "number",
      "format": "float",
      "description": "Positive values penalize tokens that have already appeared in the generated text, increasing the likelihood of generating more diverse content."
    },
    "frequencyPenalty": {
      "type": "number",
      "format": "float",
      "description": "Positive values penalize tokens based on their frequency in the generated text, decreasing the likelihood of repeating the same content verbatim."
    },
    "seed": {
      "type": "integer",
      "format": "int32",
      "description": "Seed used in decoding. If specified with the same seed and parameters, the model will attempt to return the same response."
    }
  }
}