展示HN:Comparegpt.io – 通过多模型比较识别大型语言模型的幻觉

1作者: tinatina_AI2 天前原帖
嗨,HN,我是Tina。 和你们中的许多人一样,我对大型语言模型(LLM)的幻觉感到沮丧——这些输出听起来非常自信,但实际上是虚构的。因此,我创建了CompareGPT.io,这是一个工具,可以: - 在多个LLM(如ChatGPT、Gemini、Claude、Grok等)上运行相同的查询 - 并排显示结果,以便更容易发现差异(和幻觉) - 提供统一的API,便于集成 我们刚刚开放了候补名单(并提供半价的早鸟优惠和终身计划赠送)。 我很想听听这个社区的反馈:你们认为多模型一致性是否可以成为减少法律、金融或研究领域幻觉的标准方法? 谢谢!
查看原文
Hi HN I’m Tina. Like many of you, I’ve been frustrated by LLM hallucinations — outputs that sound perfectly confident but are fabricated. So I built CompareGPT.io, a tool that: Runs the same query across multiple LLMs (ChatGPT, Gemini, Claude, Grok, etc.) Shows results side by side so discrepancies (and hallucinations) are easier to catch Offers a unified API for easier integration We just opened a waitlist (with a half-price early offer + lifetime plan giveaway). I’d love feedback from this community: do you think multi-model consistency could become a standard way to reduce hallucinations in law, finance, or research? Thanks!