I was curious whether the distribution would vary from model to model. Here are the results for 1,000 queries each for smaller models in the Gemini, Mistral, Qw

I was curious whether the distribution would vary from model to model. Here are the results for 1,000 queries each for smaller models in the Gemini, Mistral, Qwen, DeepSeek, and GLM series: https://gally.net/temp/20260525_LLM_random_numbers/index.html https://gally.net/temp/20260525_LLM_random_numbers/index.htm... This experiment cost a total of US$0.0454 through OpenRouter.