Is this like sharing Netflix?
Kinda, but it's a proper gateway with no limits on accounts or locations. Everyone gets their own login and API key, and the pool of models is shared rather than a single seat being passed around.
How do we decide on subscriptions and costs?
Each month we each put in $ we're comfortable spending and then see what we can buy with that pool of money. Think of each dollar as a voting share — the bigger your contribution, the more say in which models we subscribe to.
Can I use my own coding tools with Ken?
Yes — your API key works with any OpenAI-compatible client. That includes opencode, Claude Code, Cursor, Aider, Cherry Studio, and more. Point them at https://token.tan.gl/v1 with your key and you're set.
What models can I use?
Whatever's on the shared channel at the time — a rotating set from our upstream provider. Check the All Models list on the dashboard for what's currently available, including which ones can see images and what they cost per token.
Are my conversations private?
From each other, yes — your login and chat history are yours. From the upstream AI provider, no more or less than any other AI service: the model sees your prompts to answer them. Nothing is mined or sold, and nothing is trained on your chats.
Is there a limit on how much I can use?
There's a weekly and a per-session quota, sized to keep the pool fair for everyone. You can see your remaining quota on the dashboard. If you're running low, it resets on a timer — or chip in more next month to raise the pool.
Can kids use this?
Yes — that's part of the point. No age restrictions from the gateway side, no account limits, and the shared cost means a kid asking homework questions isn't burning through a personal subscription. Commonsense supervision still applies.
What if a model I like gets retired?
The upstream catalog churns — models come and go. We prune retired ones and add new ones as they appear. If a favourite disappears, pick the closest alternative from the list, or flag it and we'll look for a replacement.
Do I need to install anything?
No — ken.tan.gl works in any browser, with chat, search, and model selection built in. If you want to wire up an external coding tool, that takes a small setup but nothing heavy.
Who runs this and who do I talk to if something breaks?
Mark hosts and maintains it. If something's not working — a login issue, a model error, a billing question — just message him. It's a small crew, not a support ticket queue.