Today for AI
HOT RADAR
opensourceHEAT 11.6°

AI CURATED HOT STORY · 10/28/2026

Open-Source LEANN Vector Index Shrinks RAG Storage by 97%

1 reports archived1 independent sourcesupdated 10/28/2026, 05:56:27
Synthesis & Latest Updates
1 Sources Cross-Validated

Researchers introduced LEANN, an open-source vector index that reduces a 60M-chunk RAG index from 201 GB to 6 GB using pruned graphs and GPU-batched re-embedding. While it maintains high recall with minimal storage, retrieval latency is significantly higher (2.48s vs 0.05s for HNSW), making it ideal for memory-constrained environments.

LATEST/Researchers introduced LEANN, an open-source vector index that reduces a 60M-chunk RAG index from 201 GB to 6 GB using pruned graphs and GPU-batched re-embedding. While it maintains high recall with minimal storage, retrieval latency is significantly higher (2.48s vs 0.05s for HNSW), making it ideal for memory-constrained environments.

Hourly heat curve

19 fully observed hours
Now
11.6
Peak
17.306:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 1 reports · Latest first
  1. Marktechpost (@Marktechpost)T2·78 pts

    Open-Source LEANN Vector Index Shrinks RAG Storage by 97%

    • LEANN uses pruned graphs and product quantization codes, performing GPU-batched re-embedding only on search path nodes for extreme compression.
    • On a 76GB Wikipedia corpus, LEANN requires only 4GB RAM compared to 188GB for HNSW, maintaining comparable accuracy on the Natural Questions dataset.
    • The key trade-off is retrieval speed: LEANN averages 2.48 seconds per query versus 0.05 seconds for HNSW, making it unsuitable for low-latency real-time apps.
Open-Source LEANN Vector Index Shrinks RAG Storage by 97% | Today for AI