Models.
Nilux offers two models, Nilux Standard and Nilux Pro. Standard is selected when you start. Both handle everyday coding as well as larger tasks, and both have a very large context window: the amount of conversation and code the model can keep in view at once.
Current prices and the exact size of the context window are on the pricing page.
Switch the model
Type /model, choose with the arrow keys and press Enter. The new model is used from your next message, and your choice is kept for the following sessions.
You cannot switch while Nilux is answering; press Esc first or wait until it is done.
How deeply it thinks: /effort
Before answering, the model thinks the problem through. /effort sets how much:
- low: faster and cheaper. Good for routine changes.
- high: the default, a balance of depth and cost.
- max: the deepest thinking for the hardest problems. It produces more output, so it costs more.
Your choice applies to the current conversation and the helper agents it starts, and is kept for new conversations. It works while Thinking Mode is on in /config, which it is by default.
What a switch costs
Nilux keeps a recent conversation ready, which makes each next reply cheaper. Switching the model or the effort level in the middle of a long conversation makes the model read the whole conversation again at the full price, once. The /effort screen shows this cost before you confirm. In a long conversation, it is usually better to switch at the start of a new task, after /clear.
Planning with another model
In /config, Plan Model sets a separate model for plan mode. Leave it on inherit to plan with your current model.