Developer releases model-radar and model-canary CLI tools for LLM checks

On November 6, 2026, a developer posted to DEV Community's AI section introducing two small commands they run for their AI agents: model-radar and model-canary. model-radar lists the free models on OpenRouter and actually asks each newly listed one to make a tool call, logging the result line by line as PASS, FAIL, or BUSY. The author notes that many free models claim to support tool calling but then answer in prose instead. model-canary watches the paid model the agents rely on along with the credit balance, and fails over to a cheaper model or a local one when the credits run out or the model becomes unavailable. Neither command asks a model for an opinion; they measure and act. Setup takes about ten minutes, nothing needs to be pip installed, and Python 3.11 or newer is required. The guide is validated on a fresh Ubuntu machine every week, and the author recommends running both commands without credentials before adding a key. The author states that on October 8, only 4 of the 8 models probed passed the test. No follow-up coverage of the commands, and no user feedback, has appeared, so the event rests on this single report.