Prompt de avaliação de modelos de diagnóstico por imagem (IA)

Baseado no conteúdo de referência a seguir, gere um prompt útil para ferramentas de IA que produza um JSON com metadados sobre prompts de avaliação de IA em diagnóstico médico por imagem. O prompt resultante deve, ao receber qualquer conteúdo de entrada, retornar apenas um JSON com os campos: title, content, description, ai_tool, difficulty_level, rating, is_featured, tags, topic.
Ao criar esse prompt, inclua as seguintes etapas: 1) analise qual ferramenta de IA seria mais adequada entre as opções: chatgpt, claude, deepseek, gemini, copilot, perplexity, com justificativa sucinta; 2) atribua um rating entre 1 e 5 estrelas; 3) defina o nível de dificuldade entre beginner, intermediate ou advanced; 4) defina is_featured como true ou false; 5) gere uma lista de tags relevantes; 6) atribua um nome de tópico.
Use o conteúdo de referência abaixo como contexto. Conteúdo de referência: I just gave both the same image bellow and typed diagnosis with no more info https://preview.redd.it/zzot8nnq0a2g1.png?width=1080&format=png&auto=webp&s=062960891235d58013789c7678d8fb16d6eda978 Gemini 3 pro made the same mistake as I did when I first saw it and thought it was multiple myeloma or similar due to rouleaux formation. It continued to think it was multiple myeloma even after I gave the rest of the question stem ChatGPT 5.1 thinking, however, thought for 6 minutes. I never saw it take so much time! For a moment I began to think it was broken, but then it says Hairy cell leukemia – lymphocyte with circumferential hairy cytoplasmic projections. Bulls eye! He got it right even without needing the rest of the question stem. I was impressed with gemini 3 pro and thinking it was better then 5.1, but now I\'m on the fence again. Was really impressed by this great catch from chatgpt.