Your hardware. Your models. Full agent power.

Connect a local model running on your machine to everything 1K4 offers. Document editing. Code review. Parallel agents. File management. Deploys. For free.

Link your Ollama to 1K4 Lab with a single command. No credits consumed. Ever.

Start setup See setup steps

$ npx @1key4ai/bridge@latest --setup-token <GET-FROM-https://1key4ai.com/install>

Placeholder shown. Real setup token issued after sign-in on the install page.

Three steps to local models in 1K4.

No credits, no rate limits, no data leaving your machine. The local-bridge package keeps an outbound WebSocket to 1K4 so your Lab agent and bridged Claude Code sessions can call your local models alongside cloud models.

Install Ollama and pull a model

Pick any model that fits your hardware. Llama 3, Mistral, Qwen, Gemma, or anything else Ollama serves.

$ ollama pull llama3.3:70b

Run the 1K4 Bridge

A lightweight Node process that discovers your local Ollama models and forwards calls from 1K4 Lab and Claude Code.

$ npx @1key4ai/bridge@latest

Use them anywhere in 1K4

Your local models appear in Lab's model picker. Route any project, any agent task, any review, any consult to them. $0 per call.

model: "llama3.3:70b-local"

Works alongside cloud models, not instead.

Use local models for the parts that matter (privacy-sensitive code, large-context reviews, cheap reasoning) and cloud models for the parts that don't (orchestration, image generation, fast triage). The Quorum can mix both.

Privacy by defaultYour local model runs on your machine. The bridge process is open-source. Nothing leaves your network for those calls.

Quorum supports local advisorsDefine a custom advisor backed by a local model. Useful for domain-specific review on proprietary code.

Performance depends on hardwareA 70B model needs serious RAM and a GPU. Smaller models work on laptops. We don't pretend Ollama on a MacBook beats GPT-5.