{"slug": "le-chonk-cooked-on-visual-grounding", "title": "Le Chonk cooked on visual grounding", "summary": "Mistral AI announced that its Le Chonk model scored 42% on the Dense 200 visual grounding benchmark, claiming it outperforms frontier closed models at pinpointing object locations in images. Mistral AI said the model is aimed at engineering, manufacturing, and earth observation tasks, including checking mechanical parts, extracting evidence from PDFs, and scanning gigapixel satellite imagery for hard-to-find objects.", "body_md": "Mistral AI on X: \"Le Chonk cooked on visual grounding.\nPinpoints exactly where things are in an image.\n42% on Dense 200, ahead of even frontier closed models..\"\n\nVision for engineering, manufacturing, and earth observation.\nChecks mechanical parts, pulls evidence from PDFs, scans gigapixel satellite imagery for the hardest-to-find objects.\nBon appétit.", "url": "https://wpnews.pro/news/le-chonk-cooked-on-visual-grounding", "canonical_source": "https://twitter.com/MistralAI/status/2109687906751086628", "published_at": "2026-10-12 18:42:34+00:00", "updated_at": "2026-10-12 18:48:19.189118+00:00", "lang": "en", "topics": ["computer-vision", "artificial-intelligence", "ai-research"], "entities": ["Mistral AI", "Le Chonk", "Dense 200"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/le-chonk-cooked-on-visual-grounding", "markdown": "https://wpnews.pro/news/le-chonk-cooked-on-visual-grounding.md", "text": "https://wpnews.pro/news/le-chonk-cooked-on-visual-grounding.txt", "jsonld": "https://wpnews.pro/news/le-chonk-cooked-on-visual-grounding.jsonld"}}