Encoder-free VLMs gain momentum as new study tackles scaling and architecture questions
liuziwei7 · x · 2026-11-14
Encoder-free vision-language models are gaining momentum with Gemma 4 (12B), Inkling (276B/975B) and MAI-Base-2 (2.8T). A new study shared by giffmana argues existing encoder-free models are single point estimates that say nothing about scaling — which the paper addresses — and that "encoder-free or not" is a false dichotomy hiding three separate questions: separate pre-training, separate vision params, and vision params before vs. inside the decoder. Related open-source work includes BAAI's EVE, NEO and SenseNova-U.
Related event: Meta FAIR: Dedicated Vision Capacity Matters, Not Pretrained Encoders(6 posts)→
More from Multimodal
- Harvard-MIT UGG model unifies video generation, depth and camera pose in one diffusion model — burny_tech · 2026-11-14
- CapCut Web Ultra Offers a Week of Unlimited Seedream and GPT Image 2.5 Generation — JaynitMakwana · 2026-11-14
- Solv3R demo impresses: 3DGS reconstruction runs in about a minute — jonstephens85 · 2026-11-14
- Dreamina allegedly paid $82k in 'labor' for a 4-minute AI video — AIandDesign · 2026-11-14
- CapCut Adds Seedance 2.5 for 4K Video Generation in AI Lab and Video Studio — CurieuxExplorer · 2026-11-14
- BAAI's encoder-free VLM family EVE: NeurIPS spotlight, ICCV highlight, open-sourced — liuziwei7 · 2026-11-14