{"type":"post","id":"openfrontiers-benchmark-finds-llm-agents-rarely-rediscover-recently-a83554","slug":"openfrontiers-benchmark-finds-llm-agents-rarely-rediscover-recently-a83554","title":"OpenFrontiers benchmark finds LLM agents rarely rediscover recently solved math problems without web access","description":"OpenFrontiers pairs a corpus of 12,151 open research problems with a benchmark of 200 mathematical problems, then tests seven LLM agents on a curated subset.","retrieval_nugget":"OpenFrontiers pairs a corpus of 12,151 open research problems with a benchmark of 200 mathematical problems, then tests seven LLM agents on a curated subset. Without web access, the agents rarely rediscovered recently solved problems; with live web access, success mostly reflected retrieval of published solutions. That distinction matters when evaluating research agents.","published_at":"2026-11-01","updated_at":"2026-11-01","record_date":"2026-11-01","date_kind":"published_at","topics":["agents","evals"],"entities":["OpenFrontiers","Jiarui Yao","Ruida Wang","Tong Zhang","Lean 4"],"related_patterns":["ai-engineering-evaluations-are-splitting-into-task-specific"],"novelty_label":"new_fact","what_changed":"OpenFrontiers finds closed-book rediscovery rare—at most 7 of 40 resolved problems—and says live-web success mostly reflects retrieval, adding a math-specific result to prior AI research-problem benchmarks.","prior_refs":["https://newruntime.com/posts/a-team-of-35-researchers-put-together-50-1f091c/"],"source_url":"https://arxiv.org/abs/2610.37739","source_urls":["https://arxiv.org/abs/2610.37739"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:openfrontiers-benchmark-finds-llm-agents-rarely-rediscover-recently-a83554","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/openfrontiers-benchmark-finds-llm-agents-rarely-rediscover-recently-a83554/","markdown":"https://newruntime.com/posts/openfrontiers-benchmark-finds-llm-agents-rarely-rediscover-recently-a83554.md","json":"https://newruntime.com/posts/openfrontiers-benchmark-finds-llm-agents-rarely-rediscover-recently-a83554.json"},"markdown_url":"https://newruntime.com/posts/openfrontiers-benchmark-finds-llm-agents-rarely-rediscover-recently-a83554.md","next_reads":[{"type":"topic","path":"/topics/agents/","url":"https://newruntime.com/topics/agents/","title":"Agents","reason":"Explore the agents topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/agents.md","json_url":"https://newruntime.com/topics/agents.json"},{"type":"topic","path":"/topics/evals/","url":"https://newruntime.com/topics/evals/","title":"Evals","reason":"Explore the evals topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/evals.md","json_url":"https://newruntime.com/topics/evals.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}