Metadata-Version: 2.4
Name: apicrusher
Version: 2.0.1
Summary: Cut AI API costs by 63-99% with intelligent routing across all AI providers
Home-page: https://apicrusher.com
Author: APICrusher
Author-email: hello@apicrusher.com
License: MIT
Project-URL: Website, https://apicrusher.com
Project-URL: Documentation, https://apicrusher.com/docs
Project-URL: Source, https://github.com/apicrusher/apicrusher
Project-URL: Bug Reports, https://github.com/apicrusher/apicrusher/issues
Project-URL: Changelog, https://apicrusher.com/changelog
Keywords: openai,anthropic,google,ai,api,optimization,cost-reduction,llm,gpt,claude,gemini,gpt-5,gpt-4,claude-4,api-optimization,ai-cost-reduction,universal-ai,groq,mistral,cohere,deepseek,llama
Platform: any
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.7
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Development Status :: 5 - Production/Stable
Classifier: Intended Audience :: Developers
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.7
Description-Content-Type: text/markdown
Requires-Dist: requests>=2.28.0
Provides-Extra: openai
Requires-Dist: openai>=1.0.0; extra == "openai"
Provides-Extra: anthropic
Requires-Dist: anthropic>=0.25.0; extra == "anthropic"
Provides-Extra: google
Requires-Dist: google-generativeai>=0.3.0; extra == "google"
Provides-Extra: cohere
Requires-Dist: cohere>=4.0.0; extra == "cohere"
Provides-Extra: redis
Requires-Dist: redis>=4.5.0; extra == "redis"
Provides-Extra: standard
Requires-Dist: openai>=1.0.0; extra == "standard"
Requires-Dist: redis>=4.5.0; extra == "standard"
Provides-Extra: all
Requires-Dist: openai>=1.0.0; extra == "all"
Requires-Dist: anthropic>=0.25.0; extra == "all"
Requires-Dist: google-generativeai>=0.3.0; extra == "all"
Requires-Dist: cohere>=4.0.0; extra == "all"
Requires-Dist: redis>=4.5.0; extra == "all"
Provides-Extra: dev
Requires-Dist: pytest>=7.0.0; extra == "dev"
Requires-Dist: black>=22.0.0; extra == "dev"
Requires-Dist: flake8>=4.0.0; extra == "dev"
Requires-Dist: twine>=4.0.0; extra == "dev"
Requires-Dist: wheel>=0.37.0; extra == "dev"
Dynamic: author
Dynamic: author-email
Dynamic: classifier
Dynamic: description
Dynamic: description-content-type
Dynamic: home-page
Dynamic: keywords
Dynamic: license
Dynamic: platform
Dynamic: project-url
Dynamic: provides-extra
Dynamic: requires-dist
Dynamic: requires-python
Dynamic: summary

# APICrusher - Cut AI API Costs by 63-99%

Stop bleeding money on AI APIs. APICrusher automatically routes requests to the cheapest capable model and caches responses, cutting costs by 63-99% with just 2 lines of code.

## 🚀 Quick Start

```bash
# Python 3.7+ required
# Use a virtual environment (recommended for all Python packages)
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install with your provider(s)
pip install apicrusher[standard]  # OpenAI + Redis caching (recommended)
# OR
pip install apicrusher[all]       # All providers (OpenAI, Anthropic, Google, etc.)
```

## 📋 Installation Options

```bash
# Choose based on your needs:
pip install apicrusher[standard]   # OpenAI + Redis (most users)
pip install apicrusher[openai]     # Just OpenAI support
pip install apicrusher[anthropic]  # Just Anthropic support
pip install apicrusher[google]     # Just Google support
pip install apicrusher[all]        # Everything - all providers
pip install apicrusher              # Minimal - add providers later
```

### Virtual Environment Required
Modern Python systems require virtual environments for pip packages:
```bash
# If you see "externally-managed-environment" error:
python3 -m venv venv
source venv/bin/activate
pip install apicrusher[standard]
```

### API Keys Required
APICrusher optimizes your existing AI API calls. You need:
1. **Your AI provider API key** (OpenAI, Anthropic, etc.) - Keep using your existing keys
2. **An APICrusher optimization key** from [apicrusher.com](https://apicrusher.com) - Enables cost optimization

### How It Works
APICrusher is a smart proxy layer. You keep your existing API keys. We analyze each request and route it to the optimal model. Your API keys never leave your server.

## 💻 Basic Usage

```python
# Before (expensive)
from openai import OpenAI
client = OpenAI(api_key="sk-...")  # Your OpenAI key

# After (63-99% cheaper)
from apicrusher import OpenAI
client = OpenAI(
    openai_api_key="sk-...",        # Your existing OpenAI key
    apicrusher_key="apc_live_..."   # Add optimization key
)

# Your code stays exactly the same
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello!"}]
)
```

## 💰 How Much Can You Save?

### Single Provider vs Cross-Provider
- **Single Provider** (e.g., just OpenAI): 70-85% savings
- **Cross-Provider** (e.g., OpenAI + Anthropic): Up to 99% savings

| Your Current Usage | Monthly Cost | With APICrusher | You Save |
|-------------------|--------------|-----------------|----------|
| GPT-4 for everything | $1,000 | $180 | $820 (82%) |
| Mixed GPT-4/3.5 | $500 | $95 | $405 (81%) |
| Heavy API usage | $5,000 | $750 | $4,250 (85%) |
| Long conversations | $2,000 | $340 | $1,660 (83%) |

## 🎯 Core Features

### Cross-Provider Optimization (NEW in v2.0)

Get 99% savings by routing between providers:

```python
from apicrusher import OpenAI

client = OpenAI(
    openai_api_key="sk-...",             # Your OpenAI key
    anthropic_api_key="sk-ant-...",      # Optional: Add for 99% savings
    google_api_key="...",                # Optional: Add Google key
    apicrusher_key="apc_live_..."    
)

# Simple GPT-4 queries now route to Claude Haiku automatically (99% cheaper)
# Complex queries stay on GPT-4 to preserve quality
```

### Universal Provider Support
Works with ALL major AI providers and models:
- **OpenAI**: GPT-5, GPT-4, GPT-4o, O1, O3 (all current models)
- **Anthropic**: Claude Opus 4.1, Claude Sonnet 4, Claude 3.5
- **Google**: Gemini 2.0, Gemini 1.5 Pro/Flash
- **Meta**: Llama 3.3, Llama 3.2, Code Llama
- **Others**: Groq, Cohere, Mistral, DeepSeek, and more

### Intelligent Model Routing
- Simple queries → Cheap models (gpt-4o-mini)
- Complex queries → Premium models (GPT-5, Claude Opus 4.1)
- Automatic quality preservation

### Smart Caching
- Deduplicates identical requests
- Redis + in-memory fallback
- 33% average cache hit rate

### 🆕 Context Compression (NEW!)
**Stop paying to reprocess the same conversation 50 times:**

```python
# Enable context compression for long conversations
response = client.chat.completions.create(
    model="gpt-4",
    messages=conversation_history,  # 50 messages = 15,000 tokens normally
    compress_context=True  # Reduces to ~3,000 tokens automatically
)

# Features:
# - Summarizes older messages while preserving key decisions
# - Removes duplicate context automatically  
# - Compresses code blocks by 40-60%
# - Sends only deltas for continuing conversations
# - Preserves last 3 messages in full for accuracy
```

**Context Compression Savings Example:**
- Normal 20-message conversation: 150,000 tokens ($2.25)
- With compression: 35,000 tokens ($0.52)
- **Savings: 77% on long conversations**

## 🏢 Enterprise Security & Controls (v2.0)

### Security Features
- **IP Allowlisting**: Restrict API keys to specific IP ranges
- **Audit Logging**: Complete usage trail for compliance (SOC2 Type II)
- **API Key Rotation**: Rotate keys without service interruption
- **Email Verification**: Secure access control for all users
- **Role-Based Access**: Admin and user permissions for teams

### Business Controls
- **Usage Quotas**: Configurable limits per tier
  - Trial: 1,000 calls/day
  - Professional: 10,000 calls/day
  - Enterprise: Unlimited with alerts
- **80% Alerts**: Email warnings before quota exceeded
- **Overage Protection**: Prevent unexpected bills
- **Team Management**: Multiple users under one billing account
- **Monthly Reports**: Automated ROI reports for finance teams

### Reliability & Monitoring
- **Webhook Retry Queue**: Never miss critical payment events
- **Health Monitoring**: Real-time system status at `/metrics`
- **Automatic Failover**: Cross-provider redundancy
- **99.9% Uptime SLA**: For enterprise customers
- **Dedicated Support**: Priority response for business accounts

### Compliance
- **SOC2 Type II**: Security audit compliant
- **GDPR Ready**: Data processing agreements available
- **HIPAA Compatible**: With enterprise agreement
- **Self-Hosted Option**: Deploy in your own VPC for maximum control

## 📊 Analytics & Reporting

```python
# Get detailed savings report
client.print_savings_summary()

# Output:
# 💸 Total Saved: $127.43
# 📞 Total Calls: 1,432
# 💾 Cache Hit Rate: 34.2%
# ⚡ Optimization Rate: 91.3%
```

### Executive Dashboard
- Real-time cost savings visualization
- Model routing analytics
- Usage patterns and trends
- Export to CSV/Excel for finance teams
- Monthly ROI reports via email

## 🔧 Advanced Usage

### Multi-Provider Setup
```python
from apicrusher import OpenAI

# Install with: pip install apicrusher[all]
client = OpenAI(
    openai_api_key="sk-...",
    anthropic_api_key="sk-ant-...",
    google_api_key="...",
    groq_api_key="...",
    apicrusher_key="apc_..."
)

# Automatically routes to cheapest provider
response = client.chat.completions.create(
    model="gpt-4",  # Will use cheapest capable model
    messages=[{"role": "user", "content": "Format this date: 2024-01-01"}]
)
```

### Context Compression Options
```python
# Fine-tune compression behavior
response = client.chat.completions.create(
    model="gpt-4",
    messages=long_conversation,
    compress_context=True,
    compression_threshold=10,  # Start compressing after 10 messages
    preserve_recent=5  # Keep last 5 messages uncompressed
)
```

### Manual Optimization Control
```python
# Force specific model
response = client.chat.completions.create(
    model="gpt-4o-mini",  # Use this exact model
    messages=messages,
    skip_optimization=True  # Bypass routing logic
)
```

### Enterprise Configuration
```python
# Configure enterprise features
client = OpenAI(
    openai_api_key="sk-...",
    apicrusher_key="apc_enterprise_...",
    config={
        "ip_allowlist": ["192.168.1.0/24"],
        "audit_logging": True,
        "usage_quota": 50000,  # Daily limit
        "alert_threshold": 0.8,  # Alert at 80% usage
        "team_id": "eng-team-01"
    }
)
```

## 📊 Real-World Results

Based on actual customer usage:

- **E-commerce company**: Reduced costs from $8,400/mo to $1,260/mo (85% savings)
- **SaaS startup**: Cut API bills from $3,200/mo to $480/mo (85% savings)  
- **AI coding assistant**: Dropped from $12,000/mo to $2,400/mo (80% savings)
- **Customer support platform**: Saved $47,000/year with context compression
- **Data analytics firm**: 99% reduction using cross-provider routing

## 🛡️ Security & Privacy

- **Your API keys stay local** - Never sent to our servers
- **No prompt logging** - Your data remains private
- **Open source core** - Audit the optimization logic
- **SOC2 compliant** - Enterprise-ready security
- **IP allowlisting** - Restrict access to your network
- **Audit trails** - Complete usage history for compliance

## 🚀 Getting Started

1. **Install**: `pip install apicrusher[standard]`
2. **Get your key**: Sign up at [apicrusher.com](https://apicrusher.com)
3. **Add 2 lines**: Replace your import and add your key
4. **Save money**: Watch your costs drop by 63-99%

## 💰 Pricing

- **Free Trial**: 7 days, no credit card required
- **Professional**: $99/month (pays for itself in hours)
- **Enterprise**: Custom pricing for high-volume usage

Most customers save 10-50x the subscription cost in the first month.

## 🤝 Support

- Email: hello@apicrusher.com
- Documentation: [apicrusher.com/docs](https://apicrusher.com/docs)
- Issues: [github.com/apicrusher/apicrusher/issues](https://github.com/apicrusher/apicrusher/issues)
- Enterprise Support: Priority response with SLA

## License

MIT License - Use it however you want.

---

**Stop bleeding money on AI APIs. Start saving with APICrusher today.**

[Get Your Key →](https://apicrusher.com)
