Knowly

Simple, flat pricing

No per-message fees, ever. Pick a plan that fits your knowledge and traffic.

Loading plans…

Private AI, on your premises

Built-ready AI servers — keep every document in-house

Prefer not to upload anything to the cloud? Buy a Knowly appliance and run the whole assistant — chat, embeddings and reranking — inside your own building. Your data never leaves the box. Each server ships with Ubuntu and Knowly preinstalled, ready to plug in.

100% privateNo documents ever leave your network. Ideal for legal, healthcare and finance.
No per-message feesOwn the hardware, run unlimited questions. Pays for itself vs. cloud APIs.
Turnkey RAG includedThe full assistant is preinstalled and tuned — no $80K–$300K custom RAG project to fund.

Knowly AI Box — Starter

Small firms & single teams that want private AI without a data center.

Starting at$8,900

5–10 concurrent users

  • 1× NVIDIA RTX 4090 (24 GB VRAM)
  • AMD Ryzen 9 (16-core) CPU
  • 64 GB DDR5 RAM
  • 2 TB NVMe SSD
  • Runs models up to ~30B + embeddings & reranker
  • Full Knowly RAG assistant preinstalled & tuned
  • Quiet compact tower · Ubuntu preloaded
Turnkey: chat + embeddings + reranking preconfigured, plus setup & data onboarding, 1-year hardware warranty and software updates. No separate RAG build to commission.
Get a quote
Most popular
GPU GPU

Knowly AI Server — Pro

Growing companies and multi-department deployments running larger models.

Starting at$18,900

10–25 concurrent users

  • 2× NVIDIA RTX 5090 (64 GB VRAM combined)
  • AMD Threadripper / Xeon (24-core+) CPU
  • 128 GB DDR5 ECC RAM
  • 4 TB NVMe SSD (RAID 1)
  • Runs 70B models (Q4) with reranker headroom
  • Full Knowly RAG assistant preinstalled & tuned
  • Tower or 4U rackmount · Ubuntu preloaded
Turnkey private assistant preconfigured, plus setup & data onboarding, 1-year hardware warranty and priority software support. No separate RAG build to commission.
Get a quote

Knowly AI Server — Enterprise

Enterprises and regulated industries needing maximum throughput and scale.

Starting at$49,900

25–50+ concurrent users

  • 2–4× NVIDIA RTX PRO 6000 Blackwell (96 GB each)
  • Dual AMD EPYC (64-core+) CPU
  • 256–512 GB ECC RAM
  • 8 TB+ NVMe RAID with redundancy
  • Runs 70B–120B+ with continuous batching
  • Full Knowly RAG assistant + white-glove setup
  • 2U/4U rackmount · redundant PSU
Turnkey enterprise assistant, on-site/remote deployment, 3-year warranty option, and SLA-backed priority support. Replaces a $80K–$300K+ custom on-prem RAG build.
Contact sales

Prices are starting configurations in USD and reflect the current (2026) GPU and memory market, where RTX GPUs, DDR5 and NVMe have all risen sharply. Final GPU, memory and storage are sized to your models and user count and quoted per order. Unlike a bare workstation, every Knowly appliance ships with the complete private RAG assistant — chat, embeddings and reranking — preinstalled and tuned; comparable custom on-premise RAG builds run $80,000–$300,000+ in software and services alone. Talk to us for a detailed quote.