Show HN: CompareGPT – Spotting LLM Hallucinations with Multi-Model Comparison

Hi HN I’m Tina. Like many of you, I’ve been frustrated with LLM hallucinations — outputs that sound confident but are completely fabricated. Instead of trusting a single model, I built CompareGPT.io, which runs queries across multiple LLMs (ChatGPT, Gemini, Claude, Grok, and more) and shows their answers side by side. The idea is simple: make AI outputs more trustworthy again. Current features include: Catch and reduce hallucinations via multi-model comparison Unified API keys for major LLMs (no more juggling configs) Half-price early waitlist + a giveaway for a free lifetime plan I’d love feedback from this community: Is multi-model consistency a practical guardrail against hallucinations? How do you currently deal with LLM hallucinations in high-stakes use cases (law, finance, research)? Thanks!