Run local models
Install the integrated Local AI extension, use the app-managed Ollama engine, and keep compatible models in your existing local library.
Codex alternative
A Codex alternative that works with your models.
OpenCode Superapp is a native desktop workspace powered by the OpenCode harness. Run local models, connect self-hosted endpoints, or use the cloud providers you already have without adding another app subscription.
No Superapp account. No subscription. Provider charges are separate.
superapp-local · qwen3
Build, test, and operate from one place
REMOTEDesktop connected
Ask OpenCode… ↑
ListeningRealtime control · local dictation
Why OpenCode Superapp
Keep the coding workspace. Choose how the intelligence runs.
The point is not to replace one model with another. It is to keep a capable desktop harness while choosing the model, provider, location, and cost that fit each task.
Connect self-hosted models
Point OpenCode at a compatible endpoint you control and use it from the same workspace as your cloud providers.
Use the OpenCode harness
Keep OpenCode's provider ecosystem, sessions, tools, permissions, terminal, and configuration instead of tying the workspace to one model company.
Skip another subscription
The core Superapp is free and requires no account. Use your own provider access and billing, or avoid model charges entirely with local inference.
Codex vs. OpenCode Superapp
Similar job. Different center of gravity.
Codex is centered on OpenAI and ChatGPT. OpenCode Superapp is centered on the OpenCode harness and the model access you choose.
CompareCodexOpenCode Superapp
Desktop experience
The Codex coding experience inside the ChatGPT desktop app.
A native desktop workspace built around the OpenCode runtime.
Model choice
OpenAI models by default, with API and custom provider configuration available.
OpenCode-supported cloud providers, local models, and self-hosted endpoints in one workspace.
Local AI
Local providers are available through Codex configuration.
An integrated Local AI extension manages the Ollama engine and discovers your model library.
Account and pricing
Free and paid ChatGPT plans, or usage billed through an API key.
No Superapp account or subscription for the core app. Provider charges remain separate.
Harness
Codex.
The open-source OpenCode runtime and its provider ecosystem.
Product capabilities and pricing can change. This comparison was reviewed on July 29, 2026.
How it works
One workspace, three ways to run a model.
Run it on your computer
Install Local AI from Extensions and use compatible models from your local library.
Connect your endpoint
Use an OpenAI-compatible endpoint you operate and keep control of where inference runs.
Bring your provider
Use the models and billing you already have through OpenCode-supported providers.
Questions, answered
What people ask before switching.
Is OpenCode Superapp free?
Yes. The core application, Local AI, Browser Automation, Computer Use, Dictation, and Voice with your own provider are free. Remote access is an optional one-time purchase.
Can OpenCode Superapp run local models?
Yes. Its Local AI extension can install and manage an Ollama engine, discover compatible models already in your local library, and keep local inference on your computer.
Can I use self-hosted or cloud models?
Yes. OpenCode Superapp uses the OpenCode provider system, so you can configure compatible self-hosted endpoints or use providers such as OpenAI, Anthropic, Google, and xAI.
Does OpenCode Superapp require a subscription?
No Superapp account or subscription is required for the core application. Cloud providers may charge separately for the models you choose to use.
Is OpenCode Superapp made by OpenAI?
No. OpenCode Superapp is an independent product by MedHue Labs LLC. It is powered by OpenCode and is not affiliated with or endorsed by OpenAI.
Your harness. Your models. Your bill.
Try the Codex alternative that does not ask for another subscription.
Download the free desktop app, open a project, and choose local, self-hosted, or cloud inference from the same workspace.