Choosing a Model

The model control in the composer footer opens the picker: the full catalog, each row showing the model, its privacy level, context window, and capabilities. A model served at more than one level gets one row per level, so GLM 5.2 appears at Private, Confidential, and Encrypted, and gpt-oss-20b at Anonymous and Confidential.

​

Filters

The Fusion card at the top of the picker selects two or three models for one prompt. See Fusion. The catalog is live and shows exactly what your account can select. The full list is in Model List; Choosing a Model covers which model suits which work.

​

Nito Auto

Nito Auto picks the model for you. It is included in any paid plan and is the default on those plans. The picker lists its routes as Nito Auto Low, Medium, and High. See Nito Auto and Prompt Enhancement.

​

The Default Model

New chats start on your default model, set in Settings under Chat behavior. + New resets the picker to it. If the default is no longer selectable, the picker falls back to the first available model. The model you pick in a chat is remembered for that chat. GLM 5.2 at Confidential is a different row from GLM 5.2 at Private, and a chat holding Confidential or Encrypted history marks weaker rows Incompatible until you start a new chat.

​

Reasoning Effort

Models that support reasoning expose an effort control in the composer: Off, Low, Medium, or High. Reasoning is off while Fusion is active and on models that do not support it.

​

Free Plan

The Free plan selects from a small set of open-weight models, with a daily prompt allowance and a shorter context window. A paid plan opens every available open-weight and frontier model, including the Uncensored and Encrypted rows. See Free Usage.

​

Next