{"slug": "h-company-releases-neomme-a-family-of-260m-and-800m-single-tower-multimodal-that", "title": "H Company Releases NeoMME: A Family of 260M and 800M Single-Tower Multimodal Encoders That Drop the Vision Tower and Causal Decoder", "summary": "H Company released NeoMME, a family of 260M and 800M single-tower multimodal encoders that process multilingual text and raw 32×32 image patches in one Transformer without a pretrained vision tower or causal decoder. On the ViDoRe v3 benchmark, the 260M model achieved 0.523 nDCG@10, with 255× index compression and 51.3 pages per second indexing throughput on one L40S GPU.", "body_md": "We look at NeoMME, a family of 260M and 800M bidirectional encoders from H Company. Unlike ColPali-style retrievers, it processes multilingual text tokens and raw 32×32 image patches in a single Transformer, with no pretrained vision tower and no causal decoder. We cover the masked discrete-diffusion pretraining objective, the dual dense and late-interaction retrieval heads, and the ViDoRe v3 results where the 260M model reaches 0.523 nDCG@10. We also break down the 255× index compression, the 51.3 pages per second indexing throughput on one L40S, and the text-retrieval gaps the authors acknowledge.\n\nThe post [H Company Releases NeoMME: A Family of 260M and 800M Single-Tower Multimodal Encoders That Drop the Vision Tower and Causal Decoder](https://www.marktechpost.com/2026/09/06/h-company-releases-neomme-a-family-of-260m-and-800m-single-tower-multimodal-encoders-that-drop-the-vision-tower-and-causal-decoder/) appeared first on [MarkTechPost](https://www.marktechpost.com).", "url": "https://wpnews.pro/news/h-company-releases-neomme-a-family-of-260m-and-800m-single-tower-multimodal-that", "canonical_source": "https://www.marktechpost.com/2026/09/06/h-company-releases-neomme-a-family-of-260m-and-800m-single-tower-multimodal-encoders-that-drop-the-vision-tower-and-causal-decoder/", "published_at": "2026-09-06 21:06:41+00:00", "updated_at": "2026-09-07 05:26:05.349728+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-research"], "entities": ["H Company", "NeoMME", "ViDoRe v3", "L40S"], "alternates": {"html": "https://wpnews.pro/news/h-company-releases-neomme-a-family-of-260m-and-800m-single-tower-multimodal-that", "markdown": "https://wpnews.pro/news/h-company-releases-neomme-a-family-of-260m-and-800m-single-tower-multimodal-that.md", "text": "https://wpnews.pro/news/h-company-releases-neomme-a-family-of-260m-and-800m-single-tower-multimodal-that.txt", "jsonld": "https://wpnews.pro/news/h-company-releases-neomme-a-family-of-260m-and-800m-single-tower-multimodal-that.jsonld"}}