MODEL IDgoogle-gemma-4-31b

live

Gemma 4 31B

by GoogleApril 2, 2026131K context

Gemma 4 31B is Google's flagship dense open-weights model in the Gemma 4 family. It combines strong reasoning, coding performance, native function calling, multimodal understanding across text, image, and video, and a 256K context window in a 31B-parameter open model designed for local and cloud deployment.

API Options

Platform-level options for task execution and delivery.

taskType stringrequiredvalue: textInference: Identifier for the type of task being performed

taskUUID stringrequiredUUID v4: UUID v4 identifier for tracking tasks and matching async responses. Must be unique per task.

outputFormat stringdefault: TEXT

Specifies the file format of the generated output. The available values depend on the task type and the specific model's capabilities.

Allowed values1 value

webhookURL stringURI

Specifies a webhook URL where JSON responses will be sent via HTTP POST when generation tasks complete. For batch requests with multiple results, each completed item triggers a separate webhook call as it becomes available.

Learn more1 resource

Webhooks
PLATFORM

deliveryMethod stringdefault: sync

Determines how the API delivers task results.

Allowed values3 values

: Returns complete results directly in the API response.
: Returns an immediate acknowledgment with the task UUID. Poll for results using getResponse.
: Streams results token-by-token as they are generated.

Learn more1 resource

Task Polling
PLATFORM

includeCost booleandefault: false: Include task cost in the response.

includeUsage booleandefault: false: Include token usage statistics in the response.

numberResults integermin: 1max: 4default: 1: Number of results to generate. Each result uses a different seed, producing variations of the same parameters.

Inputs

Input resources for the task (images, audio, etc). These must be nested inside the inputs object.

inputs » images images array of stringsmin items: 1: Array of image inputs (UUID, URL, Data URI, or Base64).

inputs » videos videos array of stringsmin items: 1: Array of video inputs (UUID, URL, or Base64).

Core Parameters

Primary parameters that define the task output.

model stringrequiredvalue: google-gemma-4-31b

Identifier of the model to use for generation.

Learn more3 resources

seed integermin: 0max: 9223372036854776000

Random seed for reproducible generation. When not provided, a random seed is generated in the unsigned 32-bit range.

Learn more1 resource

Seed
LEARN

messages array of objectsrequiredmin items: 1

Array of chat messages forming the conversation context. The final message must use the user role.

Properties2 properties

messages » role role stringrequired: The role of the message author.
Allowed values2 values

messages » content content stringrequiredmin: 1: The text content of the message.

Settings

Technical parameters to fine-tune the inference process. These must be nested inside the settings object.

settings » systemPrompt systemPrompt stringmin: 1max: 50000: System-level instruction that guides the model's behavior and output style across the entire generation.

settings » temperature temperature floatmin: 0max: 2step: 0.01: Controls randomness in generation. Lower values produce more deterministic outputs, higher values increase variation and creativity.

settings » topP topP floatmin: 0max: 1step: 0.01: Nucleus sampling parameter that controls diversity by limiting the probability mass. Lower values make outputs more focused, higher values increase diversity.

settings » frequencyPenalty frequencyPenalty floatmin: 0max: 2step: 0.01default: 0: Penalizes tokens based on their frequency in the output so far. A value of 0.0 disables the penalty.

settings » maxTokens maxTokens integermin: 1: Maximum number of tokens to generate in the response.

settings » minP minP floatmin: 0max: 1step: 0.01default: 0: Minimum probability threshold. Tokens with probability below this value are excluded from sampling.

settings » presencePenalty presencePenalty floatmin: -2max: 2step: 0.01default: 0: Encourages the model to introduce new topics. A value of 0.0 disables the penalty.

settings » repetitionPenalty repetitionPenalty floatmin: 0max: 2step: 0.01default: 1: Penalizes tokens that have already appeared in the output. A value of 1.0 disables the penalty.

settings » stopSequences stopSequences array of stringsmin: 1: Array of sequences that will cause the model to stop generating further tokens when encountered.

settings » thinkingLevel thinkingLevel stringdefault: high: Controls the depth of internal reasoning the model performs before generating a response.
Allowed values2 values

settings » topK topK integermin: 1max: 100: Top-K sampling parameter that limits the number of highest-probability tokens considered at each step.