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.
Show HN: RightSize CLI, Find the cheapest LLM that works for your prompt | Better HN
