Free forever · Open source · Apache 2.0

One gateway. Every LLM. Anywhere.

OllaBridge unifies every model you run — local Ollama, your gaming PC, free Colab GPUs, paid OpenAI — behind one OpenAI-compatible URL. Nodes dial out, so there's nothing to port-forward. Nothing to VPN.

Install free →
terminal — install & auto-pair
$ pip install ollabridge
$ ollabridge start

 Local gateway:   http://localhost:11435/v1
 Cloud connected: https://api.ollabridge.com
 Your API key:    sk-ollabridge-xY9kL2mN…

$ curl -H "Authorization: Bearer sk-…" localhost:11435/v1/models

Two products. One ecosystem.

Run OllaBridge alone, or pair it with the Cloud to share compute across machines and devices.

Open source pip install ollabridge

OllaBridge Local

The gateway you install. OpenAI-compatible API on top of Ollama and any local runtime — with a smart router, an admin UI, and a model catalog that auto-picks the best models for your hardware.

  • OpenAI-compatible /v1/chat/completions
  • Model catalog with auto-scoring & one-click pull
  • Multi-provider routing — local, free clouds, paid APIs
  • Admin dashboard at localhost:11435
Source on GitHub →
Hosted Auto-paired on install

OllaBridge Cloud

The relay and control plane. Pair your machines, share GPUs across devices, manage users and keys — all from a single dashboard. Devices dial out via WebSocket: nothing to port-forward.

  • TV-style pairing — pair a PC in seconds with an ABCD-1234 code
  • Works behind NAT, CGNAT, and firewalls
  • Federated mesh with overflow load-balancing
  • Google OAuth, sessions, RBAC, audit logs

Connected in 60 seconds

The gateway is one command. Pairing the cloud is automatic.

1

Install on any machine

One pip install. Runs on your laptop, gaming PC, Colab, or a cloud VM.

pip install ollabridge
2

Auto-pair to the Cloud

First start opens a browser, you sign in once, and your machine dials out — no ports opened.

ollabridge start
3

Use it from anywhere

Apps point at one URL. OllaBridge routes the request to the right node, every time.

/v1/chat/completions

Built for the way you actually ship

From a single laptop to a federated fleet.

🔌

OpenAI-compatible

Drop-in for any SDK. Change one URL, keep your code.

📡

No port forwarding

Nodes dial out via WebSocket. Works behind NAT, CGNAT, café Wi-Fi.

🎚️

Smart routing

Aliases like free-best or local-private resolve to the best node available.

📚

Model catalog

Auto-discover, score, and pull the right models for your hardware.

🔐

Real security

API keys, rate limits, HMAC tokens, Google OAuth, audit logs.

🥽

VR-ready

Powers Quest avatars and HomePilot personas out of the box.

🧠

Free + paid LLMs

Route across Gemini, Groq, OpenAI, your local Ollama — with failover.

🤖

Agent-ready

Built-in MCP server lets agents manage nodes, routes, and health.

Get started in one line

Runs on macOS, Linux, Windows. Requires Python 3.10+ and Ollama installed locally.

pip install ollabridge

Free now. Built to scale later.

The gateway is open source forever. Cloud is free while we're in early access — paid tiers arrive when you actually need them.

Local

For developers, hobbyists, homelabs.

$0

Apache 2.0, forever

  • Unlimited local gateway
  • Multi-provider routing
  • Admin dashboard + model catalog
  • Community support
Install
Free in early access

Cloud

For power users and small teams.

$0 · now

Paid plans arrive later

  • Everything in Local
  • Hosted relay + auto pairing
  • Share GPUs across devices
  • Priority routing & failover
  • Generous usage during early access

Enterprise

Self-hosted, open-source, community-driven.

Open

Apache 2.0 — run it yourself

  • SSO / SAML / SCIM (community roadmap)
  • Unlimited self-hosted requests
  • Audit logs & RBAC
  • Self-hosted by design
  • Open governance & public roadmap
Interested to contribute?