K

KIOXIA AiSAQ

Developer tool

by Kioxia

Open-source approximate-nearest-neighbour search software that serves vector search directly from SSDs with little DRAM, demonstrated at 4.8 billion vectors on a single server for RAG workloads.

Find alternatives to KIOXIA AiSAQ

What it does

• Vector search
• Knowledge retrieval

Sources

Pricing

Official documentation · 23 Oct 2026

Open-source repository; the licence field is not declared on the repository page (GitHub reports NOASSERTION), so the licence is not recorded here.

Description

Press release · 23 Oct 2026

"First announced last year, KIOXIA AiSAQ(TM) open-source software technology addresses RAG scalability challenges by enabling vector search directly from SSDs, with reduced DRAM usage." (Kioxia Europe, Duesseldorf, 16 March 2026).

What it does

Press release · 23 Oct 2026

Vector search and RAG retrieval on SSD.

Sold within

Official documentation · 23 Oct 2026

Downloaded free from the repository; no parent product.

Name

Press release · 23 Oct 2026

"KIOXIA AiSAQ(TM) open-source software technology".

URL

Official documentation · 23 Oct 2026

Kioxia's own GitHub organisation (kioxia-jp); the Kioxia Europe release links to it as the download.

Deployment

Press release · 23 Oct 2026

Runs on a single server with SSD storage and GPU-accelerated indexing.

Also from Kioxia

• CM7-R Series Hardware

  Read-intensive PCIe 5.0 NVMe 2.0 enterprise SSD series (1.92TB-30.72TB, 2.5-inch and E3.S) for which Kioxia lists artificial intelligence and machine learning among the key applications.

• CM7-V Series Hardware

  Mixed-use PCIe 5.0 NVMe 2.0 enterprise SSD series (1.6TB-12.8TB, 2.5-inch and E3.S) for which Kioxia lists artificial intelligence and machine learning among the key applications.

• KIOXIA CM9-V Series Hardware

  PCIe 5.0 enterprise NVMe SSD line built on Kioxia's 8th-generation BiCS FLASH memory, used as high-throughput storage for AI training and inference data pipelines.

• CM9-R Series Hardware

  Read-intensive PCIe 5.0 NVMe 2.0 enterprise SSD series (1.92TB-61.44TB, 2.5-inch and E3.S) for which Kioxia lists artificial intelligence and machine learning among the key applications.

• LC9 Series Hardware

  High-capacity PCIe 5.0 NVMe 2.0 enterprise SSD series (30.72TB-122.88TB across 2.5-inch, E3.L and E3.S) that Kioxia positions for LLM training, RAG inference and data-lake storage in AI systems.

• NX1 Series Hardware

  E1.S PCIe 5.0 NVMe data-centre SSD series (1.92TB-15.36TB) with direct liquid-cooling support, announced by Kioxia on 2026-07-29 for accelerator-dense AI server infrastructure, with no catalogue page or availability date published yet.

Open Kioxia in the directory

Something wrong here? Send a correction — quote this product id: kioxia-aisaq.