Cloud
The models run on our infrastructure and you talk to them from the browser. You do not need a strong machine, you need an account.
Fine tuning and deployment of AI models, at no API cost
If you have the hardware, you download them and they run on your own infrastructure. If you do not, you can use them on our cloud computing service. Either way, no API cost and no provider.
On-prem
You download the weights and run them on your own computer. After the download nothing has to come through us again.
The project
We train models and publish the weights. The reason is practical more than ideological. A model you can only reach through somebody else's interface stays somebody else's: they set the price, they set the limits, and the rules about what you may ask it change from one update to the next without anyone telling you.
Anyone with a capable machine downloads the weights and stops depending on us from that moment on. Anyone without one opens the same models in a browser and runs them on our infrastructure, for as long as the free tier holds. Both roads end in the same place; the only difference is who pays for the compute.
We do not count tokens and we do not sell access. There is no paid plan waiting in a drawer either, which is usually the thing that comes next. If keeping this up ever becomes unaffordable you will read it here, before you notice it anywhere else.
Ask for access
The cloud models need an account, and we open accounts by hand. Tell us who you are and what you want to do: if it makes sense you get a username and a password.