From Pixels to Words -- Towards Native One-Vision Models at Scale
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916056410357760 |
|---|---|
| author | Diao, Haiwen Wang, Jiahao Wu, Penghao Dong, Yuhao Niu, Yuwei Zhu, Yue Cai, Zhongang Fan, Weichen Dai, Linjun Wu, Silei Zheng, Xuanyu Li, Mingxuan Zhang, Yuanhan Li, Bo Deng, Hanming Lu, Huchuan Wang, Quan Yang, Lei Lu, Lewei Lin, Dahua Liu, Ziwei |
| author_facet | Diao, Haiwen Wang, Jiahao Wu, Penghao Dong, Yuhao Niu, Yuwei Zhu, Yue Cai, Zhongang Fan, Weichen Dai, Linjun Wu, Silei Zheng, Xuanyu Li, Mingxuan Zhang, Yuanhan Li, Bo Deng, Hanming Lu, Huchuan Wang, Quan Yang, Lei Lu, Lewei Lin, Dahua Liu, Ziwei |
| contents | Current vision-language models (VLMs) typically stitch together separate image encoders and language decoders via multi-stage alignment, a modular framework that inevitably fragments pixel-level signals across frames and scatters early pixel-word interactions. In parallel, native VLMs, despite impressive performance on single images, remain largely unexplored in multi-image, video understanding, and spatial intelligence. Hence, we introduce NEO-ov, a native foundation model that learns cross-frame and pixel-word correspondence end-to-end, without any external encoders, auxiliary adapters, or post-hoc fusion. By eliminating module boundaries entirely, NEO-ov enables fine-grained and unified spatiotemporal modeling to emerge natively inside the model. Notably, NEO-ov largely narrows the gap to modular counterparts while excelling at fine-grained visual perception, validating that native "one-vision" architectures are not only feasible but competitive at scale. Beyond empirical performance, we unveil systematic architectural analyses and detailed training recipes to facilitate subsequent native multimodal modeling. Our code and models are publicly available at: https://github.com/EvolvingLMMs-Lab/NEO. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_28820 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | From Pixels to Words -- Towards Native One-Vision Models at Scale Diao, Haiwen Wang, Jiahao Wu, Penghao Dong, Yuhao Niu, Yuwei Zhu, Yue Cai, Zhongang Fan, Weichen Dai, Linjun Wu, Silei Zheng, Xuanyu Li, Mingxuan Zhang, Yuanhan Li, Bo Deng, Hanming Lu, Huchuan Wang, Quan Yang, Lei Lu, Lewei Lin, Dahua Liu, Ziwei Computer Vision and Pattern Recognition Current vision-language models (VLMs) typically stitch together separate image encoders and language decoders via multi-stage alignment, a modular framework that inevitably fragments pixel-level signals across frames and scatters early pixel-word interactions. In parallel, native VLMs, despite impressive performance on single images, remain largely unexplored in multi-image, video understanding, and spatial intelligence. Hence, we introduce NEO-ov, a native foundation model that learns cross-frame and pixel-word correspondence end-to-end, without any external encoders, auxiliary adapters, or post-hoc fusion. By eliminating module boundaries entirely, NEO-ov enables fine-grained and unified spatiotemporal modeling to emerge natively inside the model. Notably, NEO-ov largely narrows the gap to modular counterparts while excelling at fine-grained visual perception, validating that native "one-vision" architectures are not only feasible but competitive at scale. Beyond empirical performance, we unveil systematic architectural analyses and detailed training recipes to facilitate subsequent native multimodal modeling. Our code and models are publicly available at: https://github.com/EvolvingLMMs-Lab/NEO. |
| title | From Pixels to Words -- Towards Native One-Vision Models at Scale |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.28820 |