Find Your Optimal Config

Tell us your embedding dimension, retrieval window, and what you care about most — we'll recommend the best TurboQuantDB configuration from our benchmark data.

Quick Pick — Your Use Case

Pick one and we'll set the priorities for you. Or skip to tune manually below.

Step 1 — Embedding Dimension

Nearest benchmark dataset: arxiv-768 (d=768)

Step 2 — Retrieval Window (top-k)

Step 3 — What Matters Most

🎯 Recall 50%

Higher recall = correct answer reliably in the retrieval window

💾 Compression 35%

Higher compression = less disk at scale (100M vectors)

⚡ Speed 15%

Higher speed = lower query latency (p50 ms)

Weights auto-normalized to 100%

Loading benchmark data…