Hot-Swap AI Plans
Use the AI Plans You Already Pay For
Connect your Claude, Codex and OpenCode subscriptions and your own API keys. When one runs out mid-chat, the chat moves to another and keeps going in the same place.
Your Keys
Hot-Swap AI Plans
No Data Training
Desktop-FirstEvery Plan You Pay For, in One Workspace
Bring the subscriptions and keys you already have, decide where a chat goes when one runs out, and stop planning your day around a usage meter.
Bring Your Own Plans and Keys
Sign in with your Claude subscription, ChatGPT for OpenAI Codex, or OpenCode Go, and add more than one account per provider. They all sit under Connections in Settings.
Switch Mid-Chat
Pick another connection and model from the composer. Only this chat changes, and its conversation carries over to the new engine.
Keep Going Automatically
Turn on Usage Limit Failover and a chat that reaches its limit moves to the next connection with usage left, in the order you set.
Pick the Model per Chat
Each chat keeps its own model and reasoning effort. An agent can run on its own model while it works, then hand the chat back to yours.
Billed by Your Provider
Add keys from OpenRouter, OpenAI, Anthropic, Google and more. Usage on your own key or subscription is billed by your provider, not by us.
Set It Up Once in Three Steps
Connect what you already pay for, tell PandaOS where a chat may go, and get back to the work.
Connect Your Plans
Add your Claude, Codex or OpenCode subscriptions and any API keys under Settings, Models.
Set the Order
Turn on Usage Limit Failover and choose which connections a chat may move to, top to bottom.
Keep Working
When a plan runs out, the chat moves on by itself, or you pick the next one from the composer.
Stop Planning Your Day Around a Usage Meter
A usage limit in the middle of a task usually means stopping, waiting for a reset, or starting over somewhere else.
Without PandaOS
The limit decides when you stop.
- Work stops when the plan hits its limit mid-task
- Wait hours for the usage window to reset
- Copy the conversation into another tool to keep going
- Keep a separate app open for every subscription
- Explain the task again to a new model from scratch
With PandaOS
You decide where the work goes next.
- The chat moves to another plan when one runs out
- Keep going in the same chat, with its conversation carried over
- Every subscription and key in one workspace
- Choose the order your plans are used in
- Pick the model for each chat, or for each agent
Goes Further With the Rest of PandaOS
Every connection gets the same projects, the same connected apps and the same rules. Pick up where this page leaves off.
Keep Working When a Plan Runs Out
Bring the subscriptions and keys you already have into one workspace, and let the work carry on across them.
FAQ
What happens when my plan runs out in the middle of a chat?
With Usage Limit Failover on, PandaOS moves the chat to the next connection in your list that still has usage, onto a model of the same tier, as your next message goes out. With it off, PandaOS tells you the limit was reached and lets you move the chat to another connection or wait for the reset.
Does the new model know what we were doing?
Yes. When a chat moves to another engine, PandaOS hands over the recent conversation, so the new model picks up where the last one stopped. In a very long chat some earlier details may be summarized. Your project, rules, connected apps and agents stay exactly as they were.
Which subscriptions and keys can I connect?
A Claude subscription, OpenAI Codex (sign in with ChatGPT or an API key), OpenCode Go and Claude on AWS Bedrock, plus API keys from OpenRouter, OpenAI, Anthropic, Google, DeepSeek, Groq, xAI, Mistral and more. Local models and any OpenAI-compatible endpoint work too.
Is automatic failover on every plan?
Automatic failover and several accounts on the same provider come with Pro and up, including Pro on your key. On Community you can connect one account per provider and switch a chat between them by hand. Failover is off until you turn it on.
Does the chat move back when my plan resets?
No. A chat that moved stays on its new connection, and its usage is counted there. You can move it back from the model picker in the composer whenever you like.
Do you charge for usage on my own key?
No. Usage on your own subscription or key is billed by your provider rather than by us. If you would rather not bring one, Pro with usage includes a monthly allowance.
Will connecting my Claude or ChatGPT subscription get my account banned?
PandaOS runs the official Claude Code and Codex binaries on your machine. When you add a subscription, PandaOS starts that provider's own login command and hands you their sign-in page. You authenticate with the provider directly, their binary writes the credentials into its own store on your machine, and PandaOS never receives, reads or forwards the token. Requests then go from your machine to your provider, from their own client. The honest caveat: providers are still settling how subscriptions may be used outside their own interfaces, and the guidance is not final. What we rely on is that their binary does the login and their client makes the request. If a provider draws the line differently, we will do our best to let you know.