Choosing a Model

Magus gives you access to a wide range of AI models from different providers. Each model has its own strengths, and picking the right one for your task can make a noticeable difference.

What makes models different

Not all models are built for the same purpose:

• Conversation and writing: General-purpose models are fast and well-rounded. They handle everyday tasks like drafting emails, answering questions, and explaining concepts.
• Reasoning: Some models are designed to think through complex problems step by step. They tend to be slower but more accurate on difficult logic, math, or analysis tasks.
• Creative work: Certain models excel at creative writing, brainstorming, and open-ended tasks where imagination matters.
• Search-enabled models: Some models can search the web as part of their response, which is helpful when you need current information.
• Tool use: Models that support tools can perform actions like searching your memory, running code, or scheduling jobs.

Capability badges

In the model selector, you’ll see small badges on each model indicating its special capabilities:

• search: The model can look up current information on the web.
• reason: The model supports extended thinking for complex problems.
• Modality chips (like image or →image): what the model can read and produce beyond plain text.

Each entry also shows the model’s approximate cost per request and its context window size. Not all models have all capabilities; if something doesn’t work after switching models, it may not be supported by that model.

Switching models in a conversation

You can change the model at any time during a conversation. Click the model selector in the chat input area to open a list of available models. The model you pick takes effect immediately for your next message.

Switching models mid-conversation is fine. The conversation history stays intact, and the new model will have full context from everything said so far.

Setting your default models

You can set default models in your account preferences so you don’t have to pick a model every time you start a new conversation.

Magus supports three separate default models:

• Chat model: Used for all regular conversations, search mode, and reasoning mode.
• Image model: Used when you switch to Image Generation mode.
• Video model: Used when you switch to Video Generation mode.

To update your defaults, go to Settings → Preferences and use the Default models section. Changes apply to all new conversations going forward. Existing conversations keep whichever model they were using.

Using your own models

You can connect your own AI provider account and add custom models that run on your key. They appear in the model picker alongside the catalog and are billed by your provider, not against your Magus spend cap. See Your Own Models.

If a model becomes unavailable

If a model you explicitly picked can no longer be used (for example a custom model you deleted), Magus does not silently switch to another model. The message is held and you get a notice with a Reset to default and retry button, which clears the broken selection and re-sends your message with your default model.

Usage costs

Different models have different provider costs. More powerful or specialized models typically cost more per message. A model’s relative cost is shown in the model selector so you can make an informed choice.

Choosing a more efficient model is a good way to keep PAYG spend lower. Free trial usage is limited to lower-cost models, while Pay-as-you-go unlocks premium models under your monthly spend cap.