Aleria can use different AI models for different tasks. By default it picks the best one for each request, but you can choose the model yourself when you have a preference. This gives you control over speed and reasoning depth.
Leave the model on automatic unless you have a specific reason not to. Aleria picks a strong default for each request; switch models mainly when you want faster replies or deeper reasoning.
Why Aleria runs its own models
Aleria is sovereign by design. The models run inside your own boundary, in Aleria Private Cloud or on your premises, so your prompts and files never leave and are never sent to an outside provider. That is why the catalog is made up of Aleria's own models and selected open models that can run locally, rather than third-party services that would require sending your data out. See Deployment and security.
How to pick a model
- Open or start a chat.
- Use the model selector near the chat input.
- Choose the model you want.
- Type your message and send it.
Your choice applies to that conversation until you change it again.
The models you can choose
The exact list depends on your deployment, and the model selector always shows what is available to you. A typical catalog includes:
- Default: balanced quality for everyday work. This is the one Aleria uses unless you choose another.
- Fast: quicker responses, good for short drafts, summaries, and routine questions. It also handles very long inputs well.
- Experimental: the newest model. It can be more capable but may be less stable, so use it when you want to try the latest.
- Aleria is model-agnostic, so your deployment may offer other options too. You're not locked to any single provider.
Deeper thinking
Most models offer a thinking toggle. When you turn it on, the model reasons more carefully step by step before it answers. This helps on harder problems, such as a multi-part analysis or a tricky calculation, at the cost of being a little slower. Turn it off for quick back-and-forth.
When to switch models
Different models suit different jobs:
- Speed: use the Fast model for quick answers and high-volume, routine work.
- Depth: use the Default or Experimental model, with thinking turned on, for multi-step reasoning, legal analysis, or detailed financial work.
- Visual content: when your question is about images, screenshots, or scanned documents, Aleria uses a model suited to that.
For example:
- "Draft a one-paragraph email" works well on the Fast model.
- "Review this 50-page contract for liability issues" is better on the Default model with thinking on.
- "What is wrong in this screenshot of our dashboard?" uses the visual model.
A consistent default for your team
Your administrator may set a standard model for the workspace so everyone starts from the same place, for example to meet an internal policy. When that is set, you can usually still switch models in your own chats unless the administrator has locked the choice. Ask your workspace admin if you are unsure what is configured.
Bring your own model
On on-premises deployments, your organization can bring its own model. This fits teams with specific requirements about which model weights they run. Your account team can help set this up. See Deployment and security.
Model choice and data privacy
No matter which model you pick, Aleria does not train on your data. Your prompts and files are used only to answer your questions, and they stay inside your boundary. For more details, see Deployment and security.