usefy 2.1.0

Official Python SDK for Usefy - AI Cost Control & Budget Management

Usefy Python SDK

Official Python SDK for Usefy - AI Cost Control & Budget Management.

Installation

pip install usefy

Or install from source:

pip install git+https://github.com/sherlocq61/usefy.git#subdirectory=sdk/python

Quick Start

from usefy import Usefy

# Initialize client
client = Usefy(api_key="us_live_your_key_here")

# Use like OpenAI - model auto-detection included
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello!"}]
)

print(response["choices"][0]["message"]["content"])

Features

OpenAI-Compatible Interface

# GPT-4
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Explain quantum computing"}]
)

# Claude (auto-detected)
response = client.chat.completions.create(
    model="claude-3-opus",
    messages=[{"role": "user", "content": "Write a poem"}]
)

# Gemini
response = client.chat.completions.create(
    model="gemini-2.0-flash",
    provider="google",  # Explicit provider
    messages=[{"role": "user", "content": "Hello!"}]
)

Fail-Open Mode

For maximum reliability, enable fail-open mode. If Usefy is temporarily unavailable, the SDK will signal that you should fall back to direct provider calls:

client = Usefy(
    api_key="us_live_xxx",
    fail_open=True,
    fail_open_timeout=2.0  # 2 seconds
)

try:
    response = client.chat.completions.create(
        model="gpt-4",
        messages=[{"role": "user", "content": "Hello!"}]
    )
except UsefyTimeoutError:
    if client.is_fail_open_active:
        # Fall back to direct OpenAI call
        import openai
        response = openai.chat.completions.create(...)

Pre-flight Check

Check if a request would be allowed before making it:

check = client.check(
    provider="openai",
    model="gpt-4",
    estimated_tokens=1000
)

if check["allowed"]:
    # Proceed with request
    response = client.chat.completions.create(...)
else:
    print(f"Request would be blocked: {check['reason']}")

Usage Monitoring

usage = client.get_usage()

print(f"Requests: {usage['monthly_requests']['used']}/{usage['monthly_requests']['limit']}")
print(f"Plan: {usage['plan_name']}")

Configuration

Environment Variables

export USEFY_API_KEY="us_live_your_key"

Then:

client = Usefy()  # API key from env

Full Configuration

from usefy import Usefy, UsefyConfig

config = UsefyConfig(
    api_key="us_live_xxx",
    base_url="https://api.usefy.ai",
    timeout=30.0,
    max_retries=3,
    retry_delay=1.0,
    fail_open=False,
    fail_open_timeout=2.0,
    log_level="WARNING"
)

client = Usefy(config=config)

Supported Providers

Provider	Models	Auto-Detection
OpenAI	gpt-4, gpt-3.5-turbo, o1, etc.	✅
Anthropic	claude-3-opus, claude-3-sonnet	✅
Google	gemini-2.0-flash, gemini-pro	✅
Mistral	mistral-large, mixtral	✅
Cohere	command-r, command	✅
DeepSeek	deepseek-chat	✅
Groq	llama-3.1-70b	✅
Together	various	Manual
xAI	grok-2	✅

Error Handling

from usefy import (
    Usefy,
    UsefyError,
    UsefyAuthError,
    UsefyBudgetExceeded,
    UsefyRateLimited
)

try:
    response = client.chat.completions.create(...)
except UsefyAuthError:
    print("Invalid API key")
except UsefyBudgetExceeded as e:
    print(f"Budget exceeded: {e.budget_info}")
except UsefyRateLimited as e:
    print(f"Rate limited, retry after {e.retry_after}s")
except UsefyError as e:
    print(f"Error: {e.message}")

License

MIT License - see LICENSE for details.

Support

• Documentation: https://usefy.ai/docs
• Email: [e-mail]
• Issues: https://github.com/sherlocq61/usefy/issues