Yenisei CLIFeatures
Choosing a model
Which models are available to the assistant, how they differ and how to choose a model for the conversation, a project or a single run.
The model determines the speed, depth and price of the assistant's answers. The program takes the list of models from the server and shows only those that can work as an assistant, that is, read files and run commands. The coding model is selected by default; you can change it permanently, for one project or for a single run.
Available models
| Model | What for |
|---|---|
yenisei-code | The coding model. Selected by default. |
yenisei-100 | The everyday working model. |
yenisei-max | The top-tier model for complex multi-step tasks. |
yenisei-free | A free model to try things out. |
Descriptions and prices are in the model catalog.
Changing the model
- In the conversation. The
/modelcommand shows the current model and the list of available ones;/model yenisei-maxswitches to another. The choice is remembered. - For one project. Set the model in the
modelfield of the.yenisei.jsonfile, see Settings. - For a single run. Set an environment variable before the command:
YENISEI_MODEL=yenisei-100 yenisei.
If the model is set in several ways, the variable takes precedence over the
project file, and the project file takes precedence over the choice made with
/model.
If the model is unavailable
If the model is unavailable, the program shows the server's message and suggests changing it: "Change the model: /model". Other program messages are covered on the Limits and errors page.