Empirical RAG Model Evaluation · Ask HomeWord

7-Model Quality, Hallucination & Latency Benchmark

Controlled apples-to-apples comparison across all 32 pre-populated Marriage & Parenting starter questions evaluating Gemini 3.8 Flash, Gemini 4 Carbon, Gemini 4 Argon, ChatGPT-6 Astra, ChatGPT-6.1 Sol, Claude Opus 5.5, and OpenRouter Free on identical HomeWord RAG evidence.