llm-models

llm models [options] [command]llm models list [options]llm models default [model]llm models options {list|show model|set model key value|clear model [key]}

llm models is how llm shows which model ids you can pass to llm -m / --model. Out of the box the list is OpenAI chat models. llm install plugins add vendors and local runtimes (Claude, Gemini, Ollama, llama.cpp, and others). Aliases are managed separately with llm aliases.The default model is used when a prompt has no -m. Official docs currently default that to an OpenAI chat model; llm models default changes it.llm models options stores per-model defaults so you do not repeat -o key value on every prompt.

list

Print available models (built-in OpenAI plus plugins). Default subcommand (llm models).

Keep models whose id or metadata match. Repeatable.

Restrict to specific model ids. Repeatable.

Include per-model option names (temperature, and so on).

Only models that can call tools.

Only models that support structured output schemas.

List async model implementations.

JSON instead of text.

With no argument, print the current default. With model (id or alias), store that default for llm when -m is omitted.

Default option values set for every model.

Defaults for one model.

Persist a default option (example: temperature 0.5).

Clear one option, or all defaults for that model.

Help for llm models or a subcommand.

Subcommand of llm. A model in the list still needs a working key or local backend. Plugin model ids are not available until the plugin is installed. --query matching is substring-style across the list output, not a full metadata search API.

llm(1), llm-keys(1), llm-logs(1), ollama(1), chatgpt(1)

Source code · Homepage · Documentation