{"slug": "append-filler-tokens-answer-harder-questions", "title": "Append filler tokens, answer harder questions", "summary": "Researchers found that appending 300 filler tokens (dots) to multi-hop reasoning questions significantly improves the accuracy of frontier large language models, enabling them to answer complex queries they previously failed on.", "body_md": "If you ask a frontier LLM a multi-hop reasoning question, e.g., \"Who won the Nobel Prize for Chemistry in (1900 + Mozart's age when he died)?\", it usually can't answer correctly immediately (no thinking)\nBUT if you ask the same question & append 300 dots, suddenly it can answer?", "url": "https://wpnews.pro/news/append-filler-tokens-answer-harder-questions", "canonical_source": "https://twitter.com/kaleybrauer/status/2078185882926846044", "published_at": "2026-07-18 20:38:10+00:00", "updated_at": "2026-07-18 20:51:15.274710+00:00", "lang": "en", "topics": ["large-language-models", "artificial-intelligence", "ai-research"], "entities": [], "alternates": {"html": "https://wpnews.pro/news/append-filler-tokens-answer-harder-questions", "markdown": "https://wpnews.pro/news/append-filler-tokens-answer-harder-questions.md", "text": "https://wpnews.pro/news/append-filler-tokens-answer-harder-questions.txt", "jsonld": "https://wpnews.pro/news/append-filler-tokens-answer-harder-questions.jsonld"}}