config

ownllm config get <key>
ownllm config set <key> <value>

User config is plain TOML at one of:

The file is created on first run with safe defaults; you almost never need to edit it by hand.

Recognised keys

update_channel       = "stable"   # or "beta"
auto_update          = false      # nightly self-update on the chosen channel
default_keep_alive   = "5m"       # global Ollama keep-alive override
gateway_port         = 11435      # the local OpenAI proxy listens here
api_base_url         = "https://ownllm.app"
log_level            = "info"     # one of: trace, debug, info, warn, error

All keys are optional. Anything missing falls back to the documented defaults.

Examples

ownllm config get update_channel
# stable

ownllm config set update_channel beta
ownllm config set log_level debug

The set command validates the value before writing — invalid keys exit with code 2 and a typed error message.

Per-model overrides

Per-model options (keep_alive, thinking, num_parallel) live in ~/.config/ownllm/models.toml rather than config.toml, so the two files don't fight when you reinstall a model. Use ownllm models config to edit them — don't hand- write that file unless you know the schema.

Environment variables

A few keys can be overridden by env vars (useful in containers and CI):

Env vars win over the TOML file. ownllm config get <key> reports the effective value, not just what's in the file.