Jan

Jan's desktop app and its command-line agent, Jan Agent, both take a custom provider.

Connect (desktop)

Settings → Model Providers → Add Provider:

API format: OpenAI-compatible, or Anthropic-compatible for the Messages API
Base URL: http://127.0.0.1:4010/v1 (OpenAI-compatible) or http://127.0.0.1:4010 (Anthropic-compatible; Jan adds /v1/messages)
API key: your plugboard key
Models: add each name by hand, such as fast, if the list doesn't fill on save

Turn on Tools and Vision per model if you need them; Jan doesn't detect them for custom providers.

PLUGBOARD_URL is http://127.0.0.1:4010/v1 and PLUGBOARD_KEY_JAN is your plugboard key.

jan config set \
  --provider plugboard \
  --base-url "$PLUGBOARD_URL" \
  --api-type openai \
  --model fast \
  --api-key "$PLUGBOARD_KEY_JAN"

For the Anthropic format use --api-type anthropic and the base URL without /v1. Per-project settings go in that project's agent.toml:

toml

[provider]
name = "plugboard"
base_url = "http://127.0.0.1:4010/v1"
models = ["fast"]

Both files are in examples/jan/.

Streamed chat and tool calls on both API formats.