Skip to main content

System status

Coverage is stale.

Collection is paused. Latest public event: Aug 21, 2026 (10 days ago).

← Intel index

This coverage is stale.

Last updated May 17, 2026 (about 4 months ago).

product · May 17, 2026

Qdrant Releases Version 1.18 with TurboQuant Quantization Method

Share the canonical public link.

Share as image

Qdrant released version 1.18 on May 13, 2026, introducing TurboQuant, a rotation-based vector quantization method from Google Research. Ivan Pleshkov and Jonas Schulz detailed extensions including length renormalization and per-coordinate calibration for production embeddings. The method supports 8x to 32x compression levels with benchmarks showing TQ 4-bit recall within ±2 percentage points of scalar quantization on 9 of 10 datasets and TQ 2-bit outperforming binary quantization by 9–24 percentage points on average. Qdrant 1.18 ships TurboQuant in the standard Docker image and Qdrant Cloud with migration via config change and re-index.

Failed after 3 attempts. Last error: Service unavailable

Supporting evidence