llm-openzoo

LLM plugin for models hosted by OpenZoo — an OpenAI-compatible inference endpoint with no signup: any API key value works and billing happens per call via x402 or card.

Installation

Install this plugin in the same environment as LLM:

llm install llm-openzoo

Usage

No API key is needed — the plugin defaults to the conventional sk-openzoo key. Models show up automatically:

llm models | grep openzoo

Run a prompt (note: completions are billed per call):

llm -m openzoo/z-ai/glm-5.3-flash "Three names for a pet zebra"

Streaming works by default; add --no-stream to disable it.

If you want to set an explicit key anyway (any value is accepted; useful for tagging your calls):

llm keys set openzoo
# or
export OPENZOO_API_KEY=sk-openzoo

Local gateway

To route through a local npx openzoo gateway instead of the hosted endpoint:

export OPENZOO_BASE_URL=http://localhost:8402/v1

Listing models

The plugin fetches the free model catalog from https://api.openzoo.fun/v1/models and caches it for one hour.

llm openzoo models            # human-readable list with pricing
llm openzoo models --free     # only free models
llm openzoo models --json     # raw JSON
llm openzoo refresh           # force-refresh the cached catalog

Development

To set up this plugin locally, first checkout the code. Then create a new virtual environment:

cd llm-openzoo
python -m venv venv
source venv/bin/activate
pip install -e '.[test]' || pip install -e . pytest pytest-httpx

Run the tests:

pytest

The live test that hits the real OpenZoo API is skipped by default; enable it with:

OPENZOO_LIVE_TESTS=1 pytest