Interesting that when Gemini runs a Deep Research report it references *literally* hundreds and hundreds of sources ChatGPT will return a Deep Research report with sometimes only 10 or 12 sources ChatGPT writes a better narrative…but on less stable scaffolding

I’ve written requirements in the prompt that ChatGPT must use at least 60 sources. Sometimes it listens. But sometimes it won’t complete the report. It simply can’t figure out how to check more than 20-30 sources. Or it gives up and doesn’t finish the task