CosmosDBIndexConfig

Distance metric to use for vector similarity search:

• 'cosine': Cosine similarity (default)
• 'euclidean': Euclidean distance
• 'dotproduct': Dot product

Vector index type:

• 'quantizedFlat': Good for smaller datasets (<50K vectors), default
• 'flat': Exact nearest neighbors (no quantization)
• 'diskANN': Recommended for large datasets (100K+ vectors)