Skip to main content

​
Overview

The types module provides the core type system for the Google Gen AI SDK. It includes Pydantic models and TypedDict definitions for all API request and response objects.

​
Importing Types

Types can be imported directly from the main package: 
from google import genai
from google.genai import types

# Import specific types
from google.genai.types import (
    Part,
    Content,
    GenerateContentConfig,
    HttpOptions,
)

​
Pydantic Models vs TypedDict

The SDK provides two ways to work with types:

​
Pydantic Models

Pydantic models provide validation, type checking, and convenient methods: 
from google.genai.types import Part, Content

# Create using Pydantic model
part = Part.from_text(text="Hello")
content = Content(parts=[part], role="user")
Benefits:
  • Runtime validation
  • Type safety
  • Helper methods and constructors
  • Serialization/deserialization

​
TypedDict

TypedDict definitions provide type hints for dictionaries: 
from google.genai.types import PartDict, ContentDict

# Create using plain dict
part_dict: PartDict = {"text": "Hello"}
content_dict: ContentDict = {
    "parts": [part_dict],
    "role": "user"
}
Benefits:
  • Lighter weight
  • More flexible
  • Compatible with JSON serialization
  • Editor autocomplete

​
Union Types

Many parameters accept union types for flexibility: 
from google.genai.types import PartOrDict, ContentOrDict

# These are equivalent
PartOrDict = Union[Part, PartDict]
ContentOrDict = Union[Content, ContentDict]

​
Key Type Categories

​
Content Types

Types for representing messages and content:
  • Content - Multi-part message content
  • Part - Individual content parts
  • UserContent - Content from the user
  • ModelContent - Content from the model

​
Configuration Types

Types for configuring API requests:

​
Enum Types

The module includes many enum types for API options: 
from google.genai.types import (
    HarmCategory,
    HarmBlockThreshold,
    FinishReason,
    FunctionCallingConfigMode,
)
All enums are case-insensitive: 
# These are equivalent
HarmBlockThreshold.BLOCK_MEDIUM_AND_ABOVE
HarmBlockThreshold.block_medium_and_above

​
Type Hierarchy

Content
├── UserContent (role="user")
├── ModelContent (role="model")
└── parts: List[Part]
    ├── text: str
    ├── inline_data: Blob
    ├── file_data: FileData
    ├── function_call: FunctionCall
    ├── function_response: FunctionResponse
    ├── executable_code: ExecutableCode
    └── code_execution_result: CodeExecutionResult

​
Common Patterns

​
Creating Content

from google.genai.types import UserContent, Part

# Simple text
content = UserContent("Hello, world!")

# Multiple parts
content = UserContent([
    "What's in this image?",
    Part.from_uri(file_uri="gs://bucket/image.jpg", mime_type="image/jpeg")
])

# Mixed content
content = UserContent([
    Part.from_text(text="Analyze this:"),
    Part.from_bytes(data=image_bytes, mime_type="image/png")
])

​
Using Configuration

from google.genai.types import GenerateContentConfig

config = GenerateContentConfig(
    temperature=0.7,
    max_output_tokens=1024,
    top_p=0.95,
    response_mime_type="application/json"
)

response = client.models.generate_content(
    model="gemini-2.0-flash",
    contents="Tell me a story",
    config=config
)

​
See Also

Build docs developers (and LLMs) love

Get started for freeTalk to us