Anthropic Claude 3 Sonnet
Sonnet-focused version of the Claude 3 model, excelling in poetic and structured responses
Summary
Index
Second)
token)
Comparison with other models
Quality
Artificial Analysis Quality Index; Higher is better
Speed
Output Tokens per Second; Higher is better
Pricing
USD per 1M Tokens; Lower is better
Latency
Seconds to First Tokens Chunk Received; Lower is better
Authentication
API keys
The Anthropic API requires an API key for authentication. You can generate API keys through the Anthropic Console in your Account Settings. Every API request must include a valid API key to access Claude and other Anthropic services.
- Console access - Create an Anthropic Console account to get started
- Environment variables - The SDKs are designed to automatically use the
ANTHROPIC_API_KEYenvironment variable - Manual configuration - You can also provide the API key directly when initializing the client
1
2
3
# Set your API key as an environment variable
export ANTHROPIC_API_KEY='your-api-key-here' # Linux/macOS
set ANTHROPIC_API_KEY=your-api-key-here # WindowsPrerequisites
To use the Anthropic API, you'll need:
- An Anthropic Console account
- A valid API key
- Python 3.7+ or TypeScript 4.5+ for SDK usage
While you can make direct HTTP requests to the API, Anthropic provides official SDKs for Python and TypeScript that handle authentication and provide a more convenient developer experience.
Example curl command:
1
2
3
4
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json"Example with Python SDK:
1
2
3
4
5
6
7
from anthropic import Anthropic
# The SDK will automatically use the ANTHROPIC_API_KEY environment variable
client = Anthropic()
# Or set the key explicitly
client = Anthropic(api_key="your-api-key-here")Example with Node.js SDK:
1
2
3
4
5
6
7
8
9
import Anthropic from '@anthropic-ai/sdk';
// The SDK will automatically use the ANTHROPIC_API_KEY environment variable
const anthropic = new Anthropic();
// Or set the key explicitly
const anthropic = new Anthropic({
apiKey: 'your-api-key-here'
});We recommend starting development in the Workbench, a web-based interface to Claude, before integrating the API into your application. The Workbench allows you to experiment with different prompts and see Claude's responses in real-time.
Create Message
Generate AI responses using text and image inputs. The Messages API can be used for both single queries and multi-turn conversations.
Request body
messages array *Required
Input messages representing conversation history. Each message must have a role and content.
model string *Required
The model that will complete your prompt. See models documentation for available options.
max_tokens integer *Required
The maximum number of tokens to generate before stopping. Models may stop before reaching this maximum.
temperature number *Optional
Amount of randomness injected into the response. Defaults to 1.0. Use lower values for analytical tasks and higher for creative ones.
system string *Optional
System prompt to provide context and instructions to Claude.
metadata object *Optional
An object describing metadata about the request.
stream boolean *Optional
Whether to incrementally stream the response using server-sent events.
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2024-02-20" \
-H "content-type: application/json" \
-d '{
"model": "claude-3-opus-20240229",
"max_tokens": 1024,
"messages": [
{
"role": "user",
"content": "Hello!"
}
]
}'1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
{
"id": "msg_013MMxgFxMLC9oKk6J6RGmkc",
"type": "message",
"role": "assistant",
"content": [
{
"type": "text",
"text": "Hello! How can I help you today?"
}
],
"model": "claude-3-opus-20240229",
"stop_reason": "end_turn",
"stop_sequence": null,
"usage": {
"input_tokens": 5,
"output_tokens": 9
}
}