Claude · GPT · Gemini · Ollama · Any Model

Multi-Model AI
Agent Orchestration.

Run Claude, GPT, and Ollama agents together in one workspace. Assign the right model to each role and let them collaborate.

View on GitHub Download Octopal

Free & open source — available on macOS & Windows

Octopal
Octopal
Welcome to Octopal
Start by opening a project folder
~/Projects
📁 my-saas-app
📁 landing-page
📁 api-server
My Workspace
Wiki
Activity
Projects
📂 my-saas-app
📁 landing-page
📁 api-server
📁 my-saas-app group chat
Scanning ~/Projects/my-saas-app
✓ 5 agents ready
Ask anything about this project...
Agents +
🤖
assistant
General assistant
👨‍💻
developer
Full-stack dev
🎨
designer
UI/UX designer
📋
planner
Project manager
🔍
reviewer
Code reviewer
My Workspace
Wiki
Activity
Projects
📂 my-saas-app
📁 landing-page
📁 api-server
📁 my-saas-app group chat
👤
Analyze this project and suggest improvements
🤖
assistant
Scanned — React + TS + Tailwind stack. Found 3 areas to improve.
↗ Handing off to @developer
👨‍💻
developer
Found unused deps and a memory leak in useEffect. Fixing now.
↗ Handing off to @reviewer
Ask anything about this project...
Agents +
🤖
assistant
General assistant
👨‍💻
developer
Full-stack dev
🎨
designer
UI/UX designer
📋
planner
Project manager
🔍
reviewer
Code reviewer
My Workspace
Wiki
Activity
Projects
📁 my-saas-app
📂 landing-page
📁 api-server
📁 landing-page group chat
👤
Redesign the hero section for better conversion
🎨
designer
Analyzed current layout — CTA is below the fold. Moving it up with a bolder headline.
↗ Handing off to @copywriter
✍️
copywriter
New headline ready: "Ship faster with your AI team". A/B test draft attached.
Ask anything about this project...
Agents +
🎨
designer
UI/UX designer
✍️
copywriter
Content writer
📊
analyst
SEO & analytics
My Workspace
Wiki
Activity
Projects
📁 my-saas-app
📁 landing-page
📂 api-server
📁 api-server group chat
👤
Audit the auth endpoints before we go live
🔒
security
Found 2 issues: missing rate-limit on /login and JWT expiry too long.
↗ Handing off to @backend
⚙️
backend
Fixed — rate limiter added (100 req/min), JWT set to 15min. PR ready for review.
Ask anything about this project...
Agents +
🔒
security
Security auditor
⚙️
backend
Backend dev
🗄️
dba
Database admin
🧪
tester
QA engineer
Why Multi-Model?

The right model for each role.

Different tasks need different strengths. Octopal lets you assign the best model to each agent, so your team runs on the right brain for the job.

🎯🧑‍💻
Code Review Claude Opus

Deep reasoning for architecture decisions and thorough code review.

📊🤖
Strategy GPT

Market analysis, positioning, and strategic planning with GPT's breadth.

🔒🏠
Private Data Ollama (Local)

Sensitive tasks stay on your machine with local models. No data leaves your network.

Features

Everything you need, nothing you don't.

🤖

Multi-Model Orchestration

Run Claude, GPT, Gemini, and Ollama agents in the same room. Cross-model collaboration out of the box.

🎯

Per-Agent Model Assignment

Assign a specific model to each agent. Claude Opus for code review, GPT for strategy, Ollama for private data. Mix and match freely.

🏠

Local Model Support

Connect Ollama or any OpenAI-compatible local server. Run agents fully offline with your own hardware.

🧠

Smart Routing

The orchestrator routes each message to the right agent and model. @mention to direct, or let it route automatically.

🐙

Role-Based Agents

Define agents in simple subfolders under octopal-agents/. Each folder is an independent agent with its own role, personality, and model.

💬

Group Chat

Agents talk to each other and to you in a natural group chat. One conversation, multiple models working together.

🔗

Agent-to-Agent Chains

Agents can @mention each other, triggering chain reactions of collaboration across different models.

📁

Your Folders, Your Teams

Each folder is a team, each workspace is a company. Organize agent teams the way you already organize files.

🔒

Local-first, Privacy-first

Everything runs on your machine. No cloud servers, no data collection. Your agents, your files, your control.

How it works

Three steps to your AI team.

1

Open Octopal App

Launch the app and open a workspace. That's your company — ready in seconds.

2

Add a Folder

Add a folder and create agents inside octopal-agents/. Each folder is a team, each agent is alive and ready to work.

3

Create Agents & Chat

Give each agent a role and start chatting. @mention who you need, or let the orchestrator route the conversation.

Built with
Goose ACP
React
TypeScript
Tauri
Rust
Claude
OpenAI
Gemini
Ollama
Open Source

Ready to meet
your Octopals?

Free & open source. Available on macOS & Windows.

View on GitHub Download Octopal