Model Serving Cost Optimizer

Optimize model serving costs while maintaining quality and latency SLAs.

## Current Serving Setup
{{current_setup}}

## Cost Breakdown
{{cost_breakdown}}

## SLA Requirements
- Latency P99: {{latency_sla}}ms
- Quality threshold: {{quality_threshold}}

Analyze optimization opportunities:

**Infrastructure Optimization**
- Right-sizing compute resources
- Spot/preemptible instance usage
- Multi-region cost comparison

**Model Optimization**
- Quantization options
- Distillation candidates
- Speculative decoding

**Traffic Optimization**
- Request batching
- Caching opportunities
- Traffic shaping

**Architecture Optimization**
- Model tiering
- Edge deployment
- Serverless vs dedicated

Provide:
- Cost reduction estimates
- Implementation roadmap
- Risk assessment
- Monitoring plan