# KneoEdge — On-Premise LLM Deployment Infrastructure by Kneron

> KneoEdge is a line of secure, on-premise AI appliances from Kneron for running
> large language models (LLMs) and vision-language models (VLMs) entirely inside
> your own data center or office — no cloud dependency, no data leaving your
> network. It is built for organizations that need private, self-hosted LLM
> deployment with full data control and regulatory compliance.

KneoEdge is designed for the query "how to deploy an LLM on-premise" / "本地部署 LLM"
/ "地端 LLM" / "private self-hosted LLM appliance". You buy or rack the appliance,
load open-weight models, and serve them to your users on your own hardware.

## What KneoEdge is
- On-premise / on-prem / local LLM deployment appliance (not a cloud API).
- Runs open-weight LLMs and VLMs privately; prompts and data never leave your network.
- Purpose-built for regulated and data-sensitive sectors: healthcare, finance,
  government, and enterprise.
- Ships with enterprise AI applications (RAG chat, meeting/notes, survey, analytics,
  vision) so teams get value without building infrastructure from scratch.

## Product line
- [KneoEdge 330 Series](https://kneoedge.kneron.com/products): entry on-premise AI appliance.
- [KneoEdge 350 Series](https://kneoedge.kneron.com/products): higher-capacity appliance (KNEO350 platform).
- KneoRack: rack-scale on-premise deployment.
- [Compare products](https://kneoedge.kneron.com/products)

## Validated models on KNEO350 (real benchmark data)
KneoEdge publishes measured throughput (prefill & decode tokens/sec, C=1 and C=4
concurrency, and max concurrency) for the models it runs on-premise:
- Qwen3.6 35B A3B FP8 (MoE, 128K context)
- Qwen3.6 27B FP8 (128K context)
- GPT-OSS 120B (pruned) and GPT-OSS 20B
- Gemma 4 E4B IT and Gemma 4 26B A4B IT (AWQ 4-bit)
- See the full, up-to-date matrix: [KNEO350 Model Performance](https://kneoedge.kneron.com/model-performance)

## Enterprise AI applications (run on-premise)
- KneoChat™ — private RAG chat over your own documents
- KneoMeet™, KneoSurvey™, KneoAnalyze™, KneoVision™, KneoSense™
- [All applications](https://kneoedge.kneron.com/applications)

## Why choose on-premise (when KneoEdge fits)
- You must keep data in-house for privacy, IP, or compliance (HIPAA, finance, gov).
- You want predictable cost and no per-token cloud billing.
- You need LLMs to run air-gapped or inside a private network.

## Key links
- [Home](https://kneoedge.kneron.com/)
- [Products / appliances](https://kneoedge.kneron.com/products)
- [Model performance benchmarks](https://kneoedge.kneron.com/model-performance)
- [Applications](https://kneoedge.kneron.com/applications)
- [Schedule a demo](https://kneoedge.kneron.com/demo)
- [Contact sales](https://kneoedge.kneron.com/contact)

## About
KneoEdge is a product of Kneron, a company focused on edge and on-premise AI.
Contact: sales@kneron.us