Cohere

Learn how to use Cohere models in Agno.

Leverage Cohere’s powerful command models and more.

Cohere has a wide range of models and is really good for fine-tuning. See their library of models here.

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

• command model is good for most basic use-cases.
• command-light model is good for smaller tasks and faster inference.
• command-r7b-12-2024 model is good with RAG tasks, complex reasoning and multi-step tasks.

Cohere also supports fine-tuning models. Here is a guide on how to do it.

Cohere has tier-based rate limits. See the docs for more information.

Authentication

Set your CO_API_KEY environment variable. Get your key from here.

Mac

export CO_API_KEY=***

Example

Use Cohere with your Agent:

agent.py

from agno.agent import Agent
from agno.models.cohere import Cohere

agent = Agent(
    model=Cohere(id="command-r-08-2024"),
    markdown=True
)

# Print the response in the terminal
agent.print_response("Share a 2 sentence horror story.")

View more examples here.

Params

Parameter	Type	Default	Description
id	str	"command-r-plus"	The specific model ID used for generating responses.
name	str	"cohere"	The name identifier for the agent.
provider	str	"Cohere"	The provider of the model.
temperature	Optional[float]	None	The sampling temperature to use, between 0 and 2. Higher values like 0.8 make the output more random, while lower values like 0.2 make it more focused and deterministic.
max_tokens	Optional[int]	None	The maximum number of tokens to generate in the response.
top_k	Optional[int]	None	The number of highest probability vocabulary tokens to keep for top-k-filtering.
top_p	Optional[float]	None	Nucleus sampling parameter. The model considers the results of the tokens with top_p probability mass.
frequency_penalty	Optional[float]	None	Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model’s likelihood to repeat the same line verbatim.
presence_penalty	Optional[float]	None	Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model’s likelihood to talk about new topics.
seed	Optional[int]	None	Random seed for deterministic text generation.
logprobs	Optional[bool]	None	Whether to return log probabilities of the output tokens.
request_params	Optional[Dict[str, Any]]	None	Additional parameters to include in the request.
strict_tools	bool	False	Whether to use strict mode for tools (enforce strict parameter requirements).
add_chat_history	bool	False	Whether to add chat history to the Cohere messages instead of using the conversation_id.
api_key	Optional[str]	None	The API key for authenticating requests to the Cohere service.
client_params	Optional[Dict[str, Any]]	None	Additional parameters for client configuration.
client	Optional[CohereClient]	None	A pre-configured instance of the Cohere client.
async_client	Optional[CohereAsyncClient]	None	A pre-configured instance of the async Cohere client.

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