One engine. Two destinations.
Courier is the AI engine we've built — high-performance inference, reliable tool calling, hallucination detection, the whole stack. We ship it two ways: Courier OS, our Mac ecosystem that bundles Courier with Scout, the MCP Store, and your own model workbench and library — or Courier Cloud, our flat-rate hosted API Platform.
Courier
The AI infrastructure software underneath everything. Inference, tool calling, analytics, MLX-native — built for Apple Silicon.
Courier OS
The Mac ecosystem. Bundles Courier with everything we ship for local AI — your hardware, your data, no subscription metering — all private and air-gapped.
- Scout — air-gapped coding agent, with Scout Code & Emissary
- API Platform — unlimited local inference
- MCP Store — install apps Scout can use instantly
- Model Workbench & Library — manage local models
Courier Cloud
The hosted API. Same Courier engine running on our Mac Studios — flat-rate pricing, zero install, no Mac required.
- API Platform — flat-rate, $0–$100/mo
- Curated Model Library — curated by us
- Scout (Mac-only)
- MCP Store (Mac-only)
- Custom Model Workbench (Mac-only)
Side by side
Pricing, hosting, and what ships in each delivery.
| Product | Courier OS | Courier Cloud |
|---|---|---|
API Platform OpenAI-compatible APIs | ✓ Unlimited Personal free · Commercial $300/mo | ✓ Flat-rate $0–$100/mo |
Scout Local coding agent + autonomous tasks | ✓ Included | — |
MCP Store Connect Claude, Cursor, n8n | ✓ Included | — |
Model Workbench What models you can run | ✓ Customizable Almanac + custom models | ✓ Curated Library set by Courier |
| Hosting | M-series Mac you run | Mac Studios we run |
Scout only ships in Courier OS because its Shadow Manager and OS agent need full local filesystem access to work safely. The same engine powers Courier OS and Courier Cloud — the API Platform on Cloud is the same software running on Courier OS.
Courier OS is built from the ground up to be self-hosted. You can rest-assured that your data is air-gapped and never sent to the cloud, unless you explicitly choose to web tunnel or use Courier Cloud within Courier OS.
AI Token Spend Increased ~33.2% in 2025
As AI adoption grows so do token costs. Companies are increasing their budgets every year to account for their AI usage and privacy concerns.
No Data Control
Your sensitive data flows through third-party servers. You have no control over security or compliance.
Unpredictable Costs
Per-token pricing eats your margins as you grow.
Vendor Lock-In
Closed models and proprietary APIs trap your business.
Three Pillars
Courier OS operates around 3 core pillars
Private
Your data stays on your Mac or in infrastructure you control — not someone else's token factory.
Performant
Production-ready from day one — fast MLX inference, robust APIs, and a polished experience built for real workloads, not demos.
Affordable
Skip unpredictable token bills with flat-rate cloud pricing or unlimited self-hosted usage on hardware you own.
No more token costs, complete privacy
With innovations from Apple and open source AI, we eliminate token-based billing with complete privacy — self-hosting your intelligence layer affordably, or scaling on our flat-rate cloud with your data under your control.
Courier Cloud
Flat-rate AI APIs starting completely free on our cloud, powered by Mac Studios.
Other Providers
- Per-token pricing that scales with you
- Expensive GPU rentals with limited compute
- Blackbox data and AI practices
$1,000s+/mo
Courier Cloud
- Starts FREE
- Flat-rate plans for production scaling
- Transparent data policies & open source AI
FREE – $100/mo
Courier OS
Instead of renting AI, you can own it. Macs are 1/10th the cost of NVIDIA GPUs and up to 50x more power efficient. Run AI on a Mac you own instead of paying for tokens and sending your data somewhere else.
Renting
When you pay for cloud solutions you are paying for three things:
- GPUs — server farms filled with expensive hardware
- AI Models — proprietary models trained by closed providers
- APIs — software that hosts models and bills you per token
Ownership
To own your AI stack, you need those same three things. Courier uses:
We chose Apple's Unified Memory Architecture because of how affordable and performant it is. Courier handles the complexity of running open source models — experiment free on our cloud, or buy a Mac and run unlimited inference on hardware you own.
Need help setting it up?
RecursionAI, the same team behind Courier, offers a professional services team to help you implement Courier OS for unlimited AI usage and total data privacy. Whether migrating from 3rd party providers or implementing AI for the first time their team can integrate start-to-finish.
Access Cutting-Edge Open-Weight Models
We've hand-selected the best open source models for local production use cases. Host any open weight model, with first-class support for our Courier Core Library models.
Complete Open-Weight Model Library
Access cutting-edge models from Hugging Face with plug-and-play deployment. Includes text generation, code, embeddings, vision, speech, and image models.
Connect Courier directly to your n8n workflows with our custom community nodes. No complex API integrations — just drag, drop, and automate.
- Native n8n credential and node support
- Auto-synced model library from your workbench
- Multi-modal support: text, vision, and audio
- Works with self-hosted Courier instances
Installing Courier community package...✔ Courier Credentials added✔ Courier LLM Node ready✔ Courier Chat Node readyReady to automate!Embed Scout in your own product
Ship a rock-solid AI agent inside your software — the Scout interface, rebranded and running locally on your users' Macs. Ready-to-go, private by default, no per-token bills.
Find the Right Mac for Your Use Case
Start from a preset or build your own flex stack, then see which Apple Silicon Mac fits your models.
API Platform
Size a production flex stack for APIs, agents, and cloud workloads — with presets for coding agents and production servers.
Open platform builderScout
Scout Code, Emissary, and chat — all local. Pick the Mac for Scout with pre-configured Gemma, embedding, and flex models calculated for personal agent workloads.
Open platform builder