Models and Credits
In Noco Factory, NocoAI runs on an AI model that you pick. The model can change from one message to the next. NocoAI and the cloud machine of each session use the credits of the workspace.
Pick a model
The model chip is in the session launcher and in the composer of a session. "The model your next message runs on. Each message can use a different one."
- Click the model chip.
- Select a model. The model marked Default is used when you do not pick one.
The menu lists the models that NocoDB manages. When the workspace has AI integrations, the menu also lists them under Your own API keys.
Use your own API keys
To run NocoAI on your own account with an AI provider, add an AI integration, for example OpenAI, Claude or Gemini, to the workspace. See Connect to an integration. The integration then shows under Your own API keys in the model menu.
What uses credits
| Usage | How NocoDB measures it |
|---|---|
| NocoAI, on a model that NocoDB manages | Per token. The rate changes with the model. |
| NocoAI, on your own API keys | No credits for AI. Your provider bills you directly. |
| The machine of a session | Machine time, for each second that the machine runs. This applies also when NocoAI uses your own API keys. |
| Building a machine | Machine time while the build runs. |
A paused machine does not use machine time. See When a machine sleeps. To stop the machine of a finished session, archive the session.
The composer shows a credit meter. When the workspace has no credits left, a session cannot start a machine or run NocoAI. Workspace owners can buy more credits. See Buying credits.
Availability
- Noco Factory is available on the paid plans of NocoDB Cloud during the beta. See Availability.
- AI integrations with your own keys are available on the Business plan and above. On the Plus plan, Your own API keys shows an upgrade badge.
Related
Last updated on