{"slug": "land-or-water", "title": "Land or Water?", "summary": "Andrej Karpathy highlighted an LLM evaluation that asks a model \"Land or Water?\" for a latitude and longitude coordinate 16,200 times and plots the answers as an image, reporting that the models know the map from compressing the internet. A commenter noted a model answering \"water\" every time scores roughly 70% because most of the planet is ocean, making Sonnet 4.5's 60% worse than never guessing land, while 90%+ is where a model \"really starts knowing the map.\" Another commenter said a model tested on the eval had a grasp of the globe on par with some of the best models from a year earlier, and that the architecture and low cost made it feasible to extract a labeled map of continents and countries.", "body_md": "Andrej Karpathy on X: \"Cool eval. Simply ask an LLM “Land or Water?” and give it a latitude and longitude coordinate as text. Ask 16,200 times, plot as image. The models know. From compressing the internet.\"\n\nCool eval. Simply ask an LLM “Land or Water?” and give it a latitude and longitude coordinate as text. Ask 16,200 times, plot as image. The models know. From compressing the internet.\n\nCool eval. Simply ask an LLM “Land or Water?” and give it a latitude and longitude coordinate as text. Ask 16,200 times, plot as image. The models know. From compressing the internet.\n\nfun detail: a model that just answers \"water\" every time scores roughly 70% here, since most of the planet is ocean. so Sonnet 4.5 at 60% is actually worse than never guessing land. 90%+ is where it really starts knowing the map\n\nremember this old post of mine? Tried it with Jev---it seems to have a grasp of the globe on par with some of the best models a year ago\ndue to the architecture & low cost, it was also feasible to extract a labeled map of continents and countries. lots of interesting details", "url": "https://wpnews.pro/news/land-or-water", "canonical_source": "https://twitter.com/karpathy/status/2105909609487872075", "published_at": "2026-10-02 07:07:16+00:00", "updated_at": "2026-10-02 07:38:17.205653+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-research"], "entities": ["Andrej Karpathy", "X", "Sonnet 4.5"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/land-or-water", "markdown": "https://wpnews.pro/news/land-or-water.md", "text": "https://wpnews.pro/news/land-or-water.txt", "jsonld": "https://wpnews.pro/news/land-or-water.jsonld"}}