Text2TextRequest

prompt stringrequired

The input text prompt to send to the model

model stringrequired

The model identifier to use for generation. See GET /v1/textcraft/models for available models

temperature number

Sampling temperature. Not supported when thinking is enabled

maxTokens number

Maximum number of output tokens to generate. Defaults to the model's configured maximum

imageUrl string

URL of an image to include as input. Only supported by vision-capable models

imageData string

Base64-encoded image data to include as input. Only supported by vision-capable models

thinking boolean

Enable extended thinking/reasoning mode. When true, the model will return a reasoning field with its internal thought process. Silently ignored for models that do not support thinking

tools

object

Tool definitions the model can call. The model will return toolCalls in the response if it decides to use a tool

property name*

object

description stringrequired

A description of what the tool does, used by the model to decide when to call it

parameters

object

required

A map of parameter names to their descriptions

property name* any

A map of parameter names to their descriptions

reasoningConfig

object

Advanced reasoning configuration for Claude models. Use thinking:true for a simpler alternative that works across all thinking-capable models

type string

Possible values: [enabled, disabled]

Whether to enable or disable extended reasoning. Only applies to Claude models on Bedrock

budgetTokens integer

Maximum number of tokens the model can use for internal reasoning. Counts against the total maxTokens budget

additionalModelRequestFields

object

Advanced: arbitrary fields passed directly to the Bedrock Converse API additionalModelRequestFields. Use thinking:true instead for thinking support

property name* any

Advanced: arbitrary fields passed directly to the Bedrock Converse API additionalModelRequestFields. Use thinking:true instead for thinking support