Molmo2: Open Video-Language AI with Pixel-Level Grounding

Kari Jaaskelainen

16 Jan 2026 — 1 min read

Most top video AIs are locked up. Molmo2 opens the door: open weights and open datasets, built to understand videos and ground that understanding by pointing to and tracking objects in the pixels.

Data you can build on: 7 new video datasets and 2 multi-image sets, including rich video captions, free-form video Q&A, complex object tracking, and a new video pointing set, all collected without closed models.
Training recipe: efficient sequence packing, message-tree encoding, bi-directional attention over vision tokens, and a novel token-weighting strategy.
Results: the 8B Molmo2 leads open models on short-video understanding, counting, and captioning, and is competitive on long videos.
Grounding wins: beats open models like Qwen3-VL on video counting (35.5 vs 29.6), and surpasses Gemini 3 Pro on some tasks (video pointing F1: 38.4 vs 20.0; video tracking J&F: 56.2 vs 41.1).

Why it matters: developers and researchers finally get transparent, high-quality building blocks for video-language systems that not only describe what is happening, but can show you where it happens. Paper: https://arxiv.org/abs/2601.10611v1

Paper: https://arxiv.org/abs/2601.10611v1

Register: https://www.AiFeta.com

AI OpenSource VisionLanguage VideoUnderstanding MachineLearning ComputerVision VLM Research Molmo2 Grounding Tracking Datasets

Meet GRACE: a moral governor for safer, more transparent AI

AI agents are getting powerful—so how do we make sure they do the right thing, not just the effective thing? Meet GRACE, a reason-based moral governor that keeps AI behavior aligned with human norms by separating moral reasoning from goal-driven decision-making. * Moral Module: uses deontic logic and explicit reasons

Why some fine-tuned LLMs miss phishing—and how to fix it

Not all fine-tuned LLMs spot phishing equally. A new study tests Llama 3.1 8B, Gemma 2 9B, and Mistral on high-stakes phishing detection—and uses SHAP and mechanistic interpretability to reveal why models do (or don’t) generalize. * Architecture × data diversity matters: Gemma 2 9B hits state-of-the-art performance (F1

AI that explains itself—by following the science

AI that explains itself—by following the science Black-box AI can be powerful, but it often can’t tell us why it made a decision. Concept Bottleneck Models (CBMs) try to fix that by predicting human-understandable concepts first, then the final answer. The catch: standard CBMs ignore domain-specific cause-and-effect and

Meet GenomAgent: A Team of AI Specialists for Smarter Genomics Q&A

TL;DR: Finding reliable answers in genomics is hard. GenomAgent turns one big AI into a coordinating team of specialists—and beats the current leader by 12% on a key benchmark. Genomic facts live across many databases. Standard chatbots struggle because they can’t flexibly query those sources. GeneGPT added

Read more

Meet GRACE: a moral governor for safer, more transparent AI

Why some fine-tuned LLMs miss phishing—and how to fix it

AI that explains itself—by following the science

Meet GenomAgent: A Team of AI Specialists for Smarter Genomics Q&A