#binary-quantization

16× smaller vectors in Qdrant: memory, recall, and latency results.

Vector search is memory-hungry. A million OpenAI embeddings is ~6 GB of RAM just for the raw floats and RAM is the expensive part of any…