Bring your own model and keys to an open source ChatGPT Work alternative
OpenAI decides which models ChatGPT Work can run, and it retires them on its own schedule. Kortix is the open source alternative that treats the model as a value you set: connect a provider with your own keys, then change the model the day a better one lands.
Set the model in the CLI and in your repo
Connect a provider with your keys, then pin a model where the work starts.
$ kortix providers login opencode-go$ kortix providers login opencode# kortix.yamltriggers: - slug: daily-digest type: cron agent: kortix model: provider/model # wire form; pins the fired session cron: '0 0 9 * * 1-5' prompt: Summarize yesterday's commits.Why the model decision belongs to you
In ChatGPT Work, OpenAI chooses which models are available and when each one retires. You select from the list your plan exposes, and a workspace administrator can narrow that list further. The model lineup is part of the product, so a new family reaches you on OpenAI's schedule, not yours. OpenAI documents the current Work models on its own model page, checked October 2026.
Kortix is the open-source AI Operating System, and it treats the model as a value you set. The value sits in your repo beside your agents and triggers, as text you can read and diff. You change it yourself, the day a better model lands, without waiting for a vendor release.
Any model provider, your keys
Kortix is model-agnostic. Any model provider with your own keys. Or the ChatGPT plan you already pay for. Or sign in with your OpenCode Console account for OpenCode Zen and Go. A connected provider key opens that provider's whole catalog, so a single project can run an Anthropic Claude-family model, an OpenAI GPT-family model, a Google Gemini-family model, or an OpenAI-compatible endpoint you host. Alongside those, Kortix's own managed models come from open-weight families such as Kimi, DeepSeek, and GLM.
Switch the day a better model lands
Because the model is a value, changing it is a small edit rather than a migration. You can set it once for the project, on an individual agent, or on a trigger that starts a session on its own.
Own the whole stack
The model choice is one part of a system you keep. Run it on Kortix Cloud, in your VPC, or on your own on-prem network. Self-host is free. The pricing page carries the managed cloud plans, and the self-hosting guide has the commands that start it on a machine you control.
Questions about models and keys.
Can I use my own API keys with Kortix?
Yes. Any model provider with your own keys. Or the ChatGPT plan you already pay for. Or sign in with your OpenCode Console account for OpenCode Zen and Go. A connected key opens that provider’s whole catalog.
Which model families can I bring?
The Claude family behind an Anthropic key, the GPT family behind an OpenAI key, the Gemini family behind a Google key, or any OpenAI-compatible endpoint of your own. Kortix-managed models come from open-weight families such as Kimi, DeepSeek, and GLM.
Is the model fixed once I set it?
No. The model is a value you set. You can choose it for the whole project, for one agent, or on a trigger that starts a session on its own, then change it when a better model lands.
Can I choose the model when I self-host?
Yes. Run it on Kortix Cloud, in your VPC, or on your own on-prem network. Self-host is free, and the model you point it at is yours to pick. The self-hosting guide has the commands.
Set the model where you own the stack.
Bring your own keys, switch families, and keep the choice in your repo.
Open source · Any model, your keys · Self-host, VPC, or on-prem