Choosing a model
Pick which AI model answers you in One, or let Auto choose for you.
One gives you access to a range of AI models. Some are fast and good for everyday questions; others are slower but stronger at reasoning, writing, or code. You can let One choose, or pick a model yourself.
Auto: the default
Every new chat starts on Auto, shown as auto in the model selector. Auto picks a suitable model for each message you send, so you don't have to choose. It's a good choice for most conversations.
Pick a model
- Select the model name at the top left of the chat, next to the One logo. This is the model selector.
- Under Models, choose the model you want. You can type in the search box to filter the list.
- Send your message. The model you picked answers it, and stays selected for the rest of that chat.
The list shows every model available in One. All plans get the same models. It refreshes on its own every few minutes, so new models appear without you having to reload.
Changing models mid-conversation is fine: the new model sees the earlier messages and carries on from there.
When you start a new chat, One switches back to Auto.
Other items in the selector
Below the models you may see:
- My Agents, which lists agents you've built and the One — Jury agent. See Agents and Jury & Judge.
Compare two models side by side
To see how two models answer the same question, select Add multi-conversation (the plus-in-a-circle icon) in the header, then pick a second model. A second answer column appears, and each message you send is answered by both.
The Jury & Judge can't be used together with a side-by-side comparison.
Presets
A preset saves a model together with its settings, so you can switch back to a setup you like in one step.
- Choose a model and adjust its settings.
- Open the Presets menu (the stacked-books icon) in the header and choose Save As Preset.
- Give it a name.
To use it later, open the presets menu and pick it. To make a preset your default for new chats, select the pin icon next to it in the menu; select the pin again to clear it. If you have a default preset, new chats start with it instead of Auto.
Model settings
Some models let you adjust settings such as temperature (how creative or predictable answers are) and maximum response length. To change them for the current chat, open the Parameters panel (the sliders icon) from the icons along the edge of the sidebar. The panel offers settings such as Temperature, Max Output Tokens, Max Context Tokens, Reasoning Effort, and custom instructions; which ones appear depends on the model. Parameters isn't available in Jury or agent chats.
How full is the conversation?
Each model can only read a certain amount of text at once. One shows how much of that space the current conversation is using with a small gauge in the message box, next to the microphone and send buttons. Hover over or tap it to see the exact figure, for example "Context window: 12,000 of 128,000 tokens used (9%)." When a conversation gets very long, start a new chat for best results.
If no models appear
If the selector shows a single item named This_user_has_no_key instead of a list of models, One couldn't load your models. Sign out and sign back in. If that doesn't help, see Signed out unexpectedly or contact support.