Client Initialization

The Client class accepts several configuration parameters to customize behavior.

Constructor Parameters

from catsu import Client

client = Client(
    verbose=False,      # Enable debug logging
    max_retries=3,      # Maximum retry attempts
    timeout=30,         # Request timeout in seconds
    api_keys={}         # Optional API keys dictionary
)

Parameters

`verbose` (bool, default: `False`)

Enable verbose logging to see detailed request/response information:

client = Client(verbose=True)

# Logs will show:
# - Request parameters
# - API endpoints called
# - Retry attempts
# - Response metadata
# - Token usage and costs

`max_retries` (int, default: `3`)

Maximum number of retry attempts for failed requests:

client = Client(max_retries=5)

# Retries are automatic for:
# - Rate limit errors (with exponential backoff)
# - Network timeouts
# - Temporary server errors (5xx)

`timeout` (int, default: `30`)

Request timeout in seconds:

client = Client(timeout=60)  # 60 second timeout

# Raises TimeoutError if request exceeds timeout

`api_keys` (dict, optional)

Override environment variable API keys:

client = Client(
    api_keys={
        "voyageai": "your-voyage-key",
        "openai": "your-openai-key",
        "cohere": "your-cohere-key",
    }
)

Provider names for api_keys:

"voyageai" - Voyage AI
"openai" - OpenAI
"cohere" - Cohere
"gemini" - Google Gemini
"jinaai" - Jina AI
"mistral" - Mistral AI
"nomic" - Nomic
"cloudflare" - Cloudflare Workers AI
"deepinfra" - DeepInfra
"mixedbread" - Mixed Bread
"togetherai" - Together AI

API Key Priority

Catsu resolves API keys in this order (highest to lowest priority):

Per-request api_key parameter

client.embed(model="voyage-3", input="Text", api_key="request-key")

Client api_keys dictionary

client = Client(api_keys={"voyageai": "client-key"})

Environment variables
```
export VOYAGE_API_KEY="env-key"
```

Complete Example

import catsu
import os

# Production configuration
client = catsu.Client(
    verbose=False,          # Disable logging in production
    max_retries=5,          # More retries for reliability
    timeout=60,             # Longer timeout for large batches
    api_keys={
        "voyageai": os.getenv("VOYAGE_KEY"),
        "openai": os.getenv("OPENAI_KEY"),
    }
)

response = client.embed(
    model="voyage-3",
    input="Production embedding request"
)

Next Steps

embed() Method - Generate embeddings synchronously
aembed() Method - Generate embeddings asynchronously
Error Handling - Handle exceptions and retries

Client Initialization

On this page