Show HN: RightSize CLI, Find the cheapest LLM that works for your prompt (opens in new tab)

RightSize runs your prompt against candidate models (Kimi, GLM, Qwen, Gemma etc.) in parallel via OpenRouter. Then it uses a stronger model as a Judge to score accuracy against a baseline model.

Happy to answer questions.