skip to content
Get models

Local models

Get models

Pull from the catalog, or import any GGUF already on your disk. Both end in the same registry.

See what you have, and what’s available

terminal
conifer models list

Installed models first (name, quantization, size, architecture, path), then the download catalog.

Pull from the catalog

terminal
conifer models pull <model-name>

Takes a catalog name from conifer models list. The download resumes if interrupted, is verified against its published SHA-256, and lands atomically. This is the only model command that downloads weights.

Import a GGUF

terminal
conifer models import <path>.gguf --name <model-name>

Registers the file in place; it is never copied. Leave --name off and the file stem becomes the name. The quantization is read from the GGUF header. Which architectures the engine runs is on Reading a model; a file it can’t run is refused with a clear error (exit 64) and nothing is installed.

Delete

terminal
conifer models remove <model-name> --yes

A pulled model’s weights are deleted from disk. An imported model is unregistered; your file stays where it was. Without --yes it asks first.