Embedding Model Benchmark Template

Design a comprehensive benchmark for evaluating embedding models for our RAG system.

## Use Case
{{use_case_description}}

## Dataset Characteristics
{{dataset_description}}

## Models to Evaluate
{{model_list}}

Create a benchmark that measures:
1. Retrieval quality (MRR, Recall@K, NDCG)
2. Inference latency (P50, P95, P99)
3. Throughput (embeddings/second)
4. Memory footprint
5. Cost per 1M tokens

Provide:
- Python code structure for the benchmark
- Recommended test dataset size
- Statistical significance testing approach
- Visualization recommendations