Cohere Python SDK

Integrate Cohere's language models into your Python applications for chat, text generation, embeddings, and more. Works across multiple platforms including AWS Bedrock, Azure, GCP, and Oracle OCI.

What This Skill Does

This skill helps you integrate the Cohere Python SDK into your project for:

Installation

Install the Cohere SDK:

```bash

pip install cohere

```

Set your API key as an environment variable (recommended):

```bash

export CO_API_KEY=your_cohere_api_key

```

Or add to your shell profile (~/.zshrc or ~/.bashrc) for persistence.

Basic Chat Usage

```python

import cohere

Initialize client (reads CO_API_KEY from environment)

co = cohere.ClientV2()

Generate a chat response

response = co.chat(

model="command-r-plus-08-2024",

messages=[{"role": "user", "content": "hello world!"}],

)

print(response)

```

Streaming Chat

For real-time streaming responses:

```python

import cohere

co = cohere.ClientV2()

response = co.chat_stream(

model="command-r-plus-08-2024",

messages=[{"role": "user", "content": "hello world!"}],

)

Stream the response token by token

for event in response:

if event.type == "content-delta":

print(event.delta.message.content.text, end="")

```

Multi-Turn Conversations

Build conversational experiences with message history:

```python

messages = [

{"role": "user", "content": "What is Python?"},

{"role": "assistant", "content": "Python is a high-level programming language..."},

{"role": "user", "content": "What are its main uses?"}

]

response = co.chat(

model="command-r-plus-08-2024",

messages=messages

)

```

Available Models

Common Cohere models:

Refer to [Cohere's documentation](https://docs.cohere.com/) for the full model list and capabilities.

Configuration

API Key Management

**Option 1: Environment Variable (Recommended)**

```bash

export CO_API_KEY=your_api_key

```

**Option 2: Explicit Initialization**

```python

co = cohere.ClientV2(api_key="your_api_key")

```

Cross-Platform Usage

The SDK supports multiple platforms. Consult the [SDK support documentation](https://docs.cohere.com/docs/cohere-works-everywhere) for platform-specific configuration (AWS Bedrock, Azure, GCP, Oracle OCI).

Best Practices

1. **Always use environment variables** for API keys - never hardcode them

2. **Use streaming** for interactive applications to reduce perceived latency

3. **Handle rate limits** - implement retry logic with exponential backoff

4. **Monitor token usage** - track API costs by monitoring response metadata

5. **Choose the right model** - balance capability vs. cost/latency for your use case

Error Handling

```python

import cohere

from cohere.errors import CohereAPIError

co = cohere.ClientV2()

try:

response = co.chat(

model="command-r-plus-08-2024",

messages=[{"role": "user", "content": "Hello"}]

)

except CohereAPIError as e:

print(f"API error: {e}")

# Handle rate limits, invalid requests, etc.

```

Documentation

Constraints