This tweet discusses a successful prompt found to prevent NanoBananaPro (a Google AI, likely Gemini) from defaulting to suggesting image generation code (Midjourney/DALL-3) when the user is looking for text-based AI output.
わたしはテキストベースAIです。MidjourneyやDALL_3でイラストを作るコードをお出ししましょうか

