Skip to content
ModelReins ModelReins Docs

ModelReins

Multi-provider AI routing on your own workers and your own keys. No third-party gateway in front of your prompts.
Get Started

What is ModelReins?

Section titled “What is ModelReins?”

ModelReins is multi-provider AI routing you own. You download the Companion, it installs a local brain called Bob on your machine, and you dispatch AI jobs across your fleet of workers. Bob learns how you work, and his brain stays on your hardware — we don’t sync it back to our servers.

The Matriarch routes every job to the right model at the right effort level. Claude for architecture, Sonnet for code, Ollama for quick tasks — all from one interface. One subscription per provider, never double-booked. When a provider rate-limits, the router fails over to the next worker on your fleet, with your keys, against your subscription.

ModelReins is a dispatch relay, not a content host. The SaaS routes metadata (who, when, which worker, what capability, what status) — it doesn’t query the content of your prompts or your AI’s responses across tenant boundaries. Your prompts don’t transit a ModelReins-owned gateway on their way to the model; workers call providers directly from your infrastructure. Your AI conversations still go to your AI provider; ModelReins routes the work, not the words.

What you get

Section titled “What you get”
  • Bob — a local brain that stores your routing patterns, memories, and preferences. Lives on your machine. We don’t have a ModelReins-side copy.
  • The Companion — desktop app with one-click install. Detects Ollama, installs it if missing, creates your account, registers your worker. Done in 3 minutes.
  • The Saddle — VSCode extension. Dispatch jobs from your editor, output streams back inline.
  • The Wall — ambient fleet dashboard. Workers, jobs, health, metrics — all at a glance.
  • The Director — routing intelligence that picks the best model for each job based on complexity, cost, and availability.

By the numbers

Section titled “By the numbers”
  • 7 providers — Claude, OpenAI, Gemini, Ollama, LM Studio, OpenRouter, 1minAI
  • 1 dashboard — monitor every worker, job, and dollar from one screen
  • $0 local jobs — Ollama and LM Studio workers cost nothing to run
  • 0 vendor lock-in — swap providers per job, per route, or per worker

Quick start

Section titled “Quick start”
  1. Download the Companion (Windows, ~90 MB)
  2. Run it — the wizard handles everything
  3. Dispatch from the Saddle or the dashboard

That’s it. Bob is on your machine. Your workers are online. The Matriarch is routing.

Providers

Compare all 7 providers — cost, speed, local vs cloud. See providers →

Cost Optimization

Route expensive jobs to cheap models. Run the rest locally for free. Optimize costs →