A new benchmark called PlaceBench has been developed to evaluate the accuracy of AI assistants in answering questions about real-world places. The benchmark consists of 150 questions across 121 cities, with answers verified against official sources. In tests conducted in November 2026, Grok Bot's AI assistant scored highest with 92 out of 100 points, demonstrating no incorrect answers. Other assistants like Muse, Claude Duet, and Instinct performed closely, while Ask Maps was the fastest but least accurate. AI
IMPACT This benchmark highlights the current limitations of AI assistants in understanding and responding to real-world place-based queries, indicating areas for improvement in factual accuracy and consistency.
RANK_REASON The item describes a new benchmark and its results for evaluating AI assistants. [lever_c_demoted from research: ic=1 ai=1.0]
- Ask Maps
- Claude Duet
- Claude Opus 5.5
- Google Maps
- Grok Bot
- Instinct
- Mexico City
- Muse
- OpenAI
- OpenStreetMap
- PlaceBench
- Slussen
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →