LangChain Reference home page

GitHub
Main Docs

Google GenAI (Gemini)

Google Vertex AI

IBM

Overview
Chat Models
LLMs
Embeddings
Rerankers

⌘I

LangChain Assistant

Ask a question to get started

Enter to send•Shift+Enter new line

Menu

Google GenAI (Gemini)

Google Vertex AI

Overview Chat Models LLMs Embeddings Rerankers

Language

Theme

Pythonlangchain-nvidia-ai-endpoints_staticsModel

Class●Since v0.0

Model

Copy

Model()

Bases

BaseModel

Attributes

Methods

View source on GitHub

model_type: Optional[Literal['chat', 'vlm', 'nv-vlm', 'embedding', 'ranking', 'completions', 'qa']]

client: Optional[Literal['ChatNVIDIA', 'NVIDIAEmbeddings', 'NVIDIARerank', 'NVIDIA']]

endpoint: Optional[str]

aliases: Optional[list]

supports_tools: Optional[bool]

supports_structured_output: Optional[bool]

supports_thinking: Optional[bool]

thinking_prefix: Optional[str]

no_thinking_prefix: Optional[str]

thinking_param_enable: Optional[dict]

thinking_param_disable: Optional[dict]

base_model: Optional[str]

validate_client

validate_thinking_config

Model information.

Thinking mode can be enabled via two mechanisms:

Tag-based: Use thinking_prefix/no_thinking_prefix (appended to system message)
Param-based: Use thinking_param_enable/thinking_param_disable (merged into request params)

All aliases are deprecated and will trigger a warning when used.

Warn if both param-based and tag-based thinking are configured.