Install

Download PandaOS and get set up.

Download and install

Go to pandaos.ai/dashboard to download PandaOS for your platform. Run the installer and launch the app.

On first launch, PandaOS checks for updates automatically and walks you through a short onboarding: your name, your expertise level (which controls how technical PandaOS communicates), and your AI provider. The onboarding is self-explanatory and takes under a minute.

Choosing your AI provider

The most important step in onboarding is choosing where PandaOS gets its AI models. Six options:

• PandaOS Cloud - the default. Your subscription includes access to models from multiple providers (Anthropic, OpenAI, Google, and more), all managed for you. Usage is metered through Credits. No API keys needed.
• Your API Keys (BYOK) - bring your own provider keys. Fifteen providers are supported, including OpenAI, Anthropic, Google (Gemini), Google Cloud (Vertex AI), Azure OpenAI, OpenRouter, DeepSeek, Groq, xAI (Grok), Mistral, Perplexity, Cohere, Fireworks AI, and Together AI, plus any OpenAI-compatible endpoint. Full control over which models you use and how billing works.
• Local - run models on your own machine. PandaOS scans for Ollama, LM Studio, KoboldCpp, and llama.cpp on their default ports and connects to whichever ones are running. You can also add an endpoint by hand for vLLM, LocalAI, Jan, or anything else that serves /v1/chat/completions. Your prompts never leave your machine.
• OpenCode Go - a subscription from opencode.ai that serves models through an OpenAI-compatible endpoint. It works with the OpenCode harness and with the default Claude Code harness.
• OpenAI Codex - connect your own ChatGPT or OpenAI account. Useful if your organization already has an OpenAI contract.
• Claude - run on your own Claude subscription, or on Claude models hosted in AWS Bedrock using a bearer token and region.

You can add or switch providers at any time in Settings, and configured providers coexist, so switching back never means re-entering keys. See Models & Harnesses for how harnesses and model selection work.