See the audit run live. Pick a traveler question and AnswerShelf queries the models right now, then shows how the answer has drifted from the fixed baseline.

Two foundation models (OpenAI GPT-4o and Anthropic Claude Sonnet 4.6) answer the question, and AnswerShelf checks each answer for Choice Hotels brands using the same deterministic detection as the static audit. The model key is held server-side and is never exposed on this page.

Reading this honestly: these are foundation models queried directly via API — their parametric prior, with no live web search; consumer assistants add retrieval and answer differently. The static audit ran four models; the live runner queries these two flagship models (same pinned versions, same detector) to bound cost, and live deltas are measured against those same two in the baseline. Choice acquired the Radisson Americas brands in 2022, so Radisson and Country Inn & Suites count as Choice.

Results will differ from the static audit, and from run to run, because model answers are non-deterministic and drift as the models update. That drift is exactly what an answer-engine visibility program has to monitor. Runs are limited and briefly cached to keep cost bounded.

Choose a question to run