RYE NANO
AI shouldn't only
live in the cloud.
Most AI depends on remote servers, subscriptions and constant connectivity. RYE NANO brings an intelligent layer directly onto your machine — reading its hardware, connecting to local models, and building an environment adapted to the computer it runs on.
Local. Adaptive.
Private. Universal.
Local
Run supported models directly on your machine.
- Local inference
- Offline operation
- Local model support
- No per-token local cost
Adaptive
RYE analyses the environment it runs inside.
- Hardware detection
- CPU / GPU / memory awareness
- Model recommendation
- Resource-aware operation
Private
Local workflows can stay local.
- Local conversations
- Local processing
- User-controlled connections
- No cloud requirement for local work
Universal
One intelligence layer across machines.
- macOS
- Windows
- Linux
One core.
Five layers.
Conversation
Natural interaction with RYE — multi-turn, streaming, interruptible.
Intelligence
Reasoning, model selection and response generation.
Work
Projects, coding, documents, research and longer tasks.
Tools
Local capabilities, integrations and system functions.
Memory & context
Project context, files and persistent intelligence where enabled.
Local when you want it.
Online when you need it.
Local
- Models running on your computer
- Offline capability
- No API inference cost
- Models you control
Hybrid — auto
- RYE chooses the path
- Based on task complexity
- And available resources
Online
- External intelligence when required
- Compatible APIs
- Search and services
- Explicitly configured by you
RYE adapts
to the machine.
RYE can evaluate operating system, memory, CPU, GPU availability, installed engines and existing local models — then recommend a configuration suited to what's actually there.
The right model
for the right job.
RYE NANO isn't one model. It's an intelligence layer above models. Different tasks need different levels of intelligence — RYE can route workloads according to complexity, available hardware and the models you've configured.
Speak.
RYE responds.
Your computer doesn't need another assistant. It needs an intelligence you can speak to — speech in, spoken answers back, continuous flow, and the ability to interrupt mid-sentence the way you would with a person.
Designed to know
what it knows.
No language model can guarantee it never produces an unsupported answer, and RYE doesn't claim to. What it is designed to do is reduce them — through confidence awareness, source-aware research, explicit uncertainty, verification of tool results, and a clear distinction between what is known and what is assumed.
And when it can't: "I don't have enough verified information to answer that reliably."
From conversation
to execution.
Work is a dedicated environment for larger projects — software, research, writing, planning, analysis and file-based tasks — while the conversation interface stays fast and simple for everything else.
Build me a Python API and explain the architecture.
Explain what is using the most memory on this machine.
Research this subject and separate verified facts from assumptions.
Debug this code and give me the corrected version.
Help me plan and write this project.
Explain this document using only the local model.
Your machine.
Your intelligence.
RYE NANO can operate locally without sending local inference requests to a cloud model. Online integrations are separate, clearly marked, and enabled only when you configure them. API keys stay under your control, and any connected service follows that service's own policy — so what stays local is stated plainly rather than promised absolutely.
From download
to intelligence.
Download RYE NANO
Open the installer
RYE analyses your machine
Choose or install the recommended engine
RYE configures itself
Start talking
RYE NANO
is the beginning.
RYE NANO Free
Intelligence for your machine.
Free downloadRYE NANO Pro
Advanced personal intelligence.
Coming soonRYE
More capable RYE systems.
In developmentMERISTEM
Intelligence for your company.
Coming laterRYE — intelligence for your machine. MERISTEM — intelligence for your company. NANO will eventually connect into MERISTEM environments; that connectivity is roadmap, not this release.
Free and Pro.
| Capability | Free | Pro |
|---|---|---|
| Local AI | Yes | Yes |
| Hardware detection | Yes | Yes |
| Model routing | Yes | Yes |
| Conversation | Yes | Yes |
| Work | Yes | Yes |
| Voice | Yes | Advanced |
| Online integrations | Basic | Advanced |
| Advanced agents | — | Yes |
| Automation | — | Yes |
| Advanced memory | — | Yes |
| MERISTEM integration | Future | Yes |
RYE NANO Free.
Version 6.3.1 — universal build. Free, no subscription.
Version
V6.3.1 — Universal
Released 31 August 2026
Downloads
—
since release
Full system requirements and release notes publish alongside documentation.
FAQ
What is RYE NANO?
A local-first AI environment that brings adaptable intelligence directly to personal computers.
Is it free?
Yes. The Free edition is available without a subscription.
Does it need internet?
Local models can operate without internet. Online features require connectivity.
Does RYE include an AI model?
RYE acts as an intelligence layer capable of working with supported local and online models.
Does RYE choose models automatically?
Where supported, RYE can recommend or route workloads according to task requirements and available hardware.
Is my data private?
Local inference can remain local. Connected online services operate according to their own configurations and policies.
Will there be a Pro version?
Yes.
Will it connect to MERISTEM?
MERISTEM connectivity is part of the broader RYE ecosystem roadmap.