← All features

Local model management

The Models sidebar tab is a full management surface for the local Ollama catalog, plus a read-only listing of the catalog reachable through your OpenAI-compatible endpoint.

The library

Every installed model gets a tile showing its family, parameter size, on-disk size, quantization and capabilities (thinking, tool use, vision). The toolbar offers:

• Pull — download a new model by tag. An inline progress chip shows percentage and current digest; you can cancel mid-pull.
• Refresh — re-query both provider catalogs.
• Search — substring match across model names.
• Open — jump into the Models editor for any installed Ollama model.

The Models editor

Opening a model loads its full configuration in a form so you can:

• Inspect the Modelfile, system prompt, chat template, parsed PARAMETER block and capability tags.
• Edit any of those fields.
• Save as… — write a new derived model with your changes. Loach never overwrites the base model, so the original is always available to revert to.
• Copy model — duplicate under a new tag without changes.
• Delete model — remove the local copy. Irreversible.
• Thinking preference — a per-model override for the Thinking toggle, sitting between the Modelfile default and any per-chat override.
• Open chat — jump straight into a fresh chat pinned to this model.

Modelfile safety

The Save as… form refuses to compile a Modelfile that would smuggle in extra directives via a malicious base tag, system block or template block. Base tags are checked against a conservative allowlist; system and template bodies are rejected if they contain delimiters that the Ollama parser would treat as an early end-of-block. You get a clear error rather than a quietly corrupted model.