Anthropic Claude

Use Anthropic Claude models with Kern agents.

Claude is a family of foundational AI models by Anthropic that can be used in a variety of applications. See their model comparisons here.

We recommend experimenting to find the best-suited model for your use-case. Here are some general recommendations:

claude-sonnet-4-20250514 model is good for most use-cases and supports image input.
claude-opus-4-1-20250805 model is their best model.
claude-3-5-haiku-20241022 model is their fastest model.

Anthropic has rate limits on their APIs. See the docs for more information.

Note

Claude API expects a max_tokens param to be sent with each request. Unless set as a param, Kern will default to 8192. See the docs for more information.

Authentication

Set your ANTHROPIC_API_KEY environment. You can get one from Anthropic here.

1export ANTHROPIC_API_KEY=***

1setx ANTHROPIC_API_KEY ***

Example

Use Claude with your Agent:

1from kern.agent import Agent
2from kern.models.anthropic import Claude
3
4agent = Agent(
5    model=Claude(id="claude-3-5-sonnet-20240620"),
6    markdown=True
7)
8
9# Print the response on the terminal
10agent.print_response("Share a 2 sentence horror story.")

Beta Features

You can use Anthropic's beta features with Kern by setting the betas parameter:

1from kern.agent import Agent
2from kern.models.anthropic import Claude
3
4agent = Agent(
5    model=Claude(
6        betas=["context-management-2025-06-27"],
7    ),
8)

Read more about beta features with Kern Claude model here.

Prompt caching

You can enable system prompt caching by setting cache_system_prompt to True:

1from kern.agent import Agent
2from kern.models.anthropic import Claude
3
4agent = Agent(
5    model=Claude(
6        id="claude-3-5-sonnet-20241022",
7        cache_system_prompt=True,
8    ),
9)

Read more about prompt caching with Kern's Claude model here.

Structured Outputs

Structured outputs are used to ensure that the model's response matches a defined schema.

This is useful to eliminate issues like missing fields or invalid values. Use it for production systems that need reliable, consistent responses in a specific format.

Kern uses Claude's native support for structured outputs. This feature is available for claude-sonnet-4-5-20250929 and all newer models. See Anthropic's structured outputs documentation for more details.

1from kern.agent import Agent
2from kern.models.anthropic import Claude
3from pydantic import BaseModel
4
5class User(BaseModel):
6    name: str
7    age: int
8    email: str
9
10agent = Agent(
11    model=Claude(id="claude-sonnet-4-5-20250929"),
12    description="Extract user information.",
13    output_schema=User,
14)

Read more about structured outputs with Kern's Claude model:

Params

Parameter	Type	Default	Description
`id`	`str`	`"claude-3-5-sonnet-20241022"`	The id of the Anthropic Claude model to use
`name`	`str`	`"Claude"`	The name of the model
`provider`	`str`	`"Anthropic"`	The provider of the model
`max_tokens`	`Optional[int]`	`4096`	Maximum number of tokens to generate in the chat completion
`thinking`	`Optional[Dict[str, Any]]`	`None`	Configuration for the thinking (reasoning) process (See their docs))
`temperature`	`Optional[float]`	`None`	Controls randomness in the model's output
`stop_sequences`	`Optional[List[str]]`	`None`	A list of strings that the model should stop generating text at
`top_p`	`Optional[float]`	`None`	Controls diversity via nucleus sampling
`top_k`	`Optional[int]`	`None`	Controls diversity via top-k sampling
`cache_system_prompt`	`Optional[bool]`	`False`	Whether to cache the system prompt for improved performance
`extended_cache_time`	`Optional[bool]`	`False`	Whether to use extended cache time (1 hour instead of default)
`request_params`	`Optional[Dict[str, Any]]`	`None`	Additional parameters to include in the request
`mcp_servers`	`Optional[List[MCPServerConfiguration]]`	`None`	List of MCP (Model Context Protocol) server configurations
`api_key`	`Optional[str]`	`None`	The API key for authenticating with Anthropic
`default_headers`	`Optional[Dict[str, Any]]`	`None`	Default headers to include in all requests
`timeout`	`Optional[float]`	`None`	Request timeout in seconds
`client_params`	`Optional[Dict[str, Any]]`	`None`	Additional parameters for client configuration
`client`	`Optional[AnthropicClient]`	`None`	A pre-configured instance of the Anthropic client
`async_client`	`Optional[AsyncAnthropicClient]`	`None`	A pre-configured instance of the async Anthropic client

Claude is a subclass of the Model class and has access to the same params.