Ai2’s Molmo 2 targets video tracking and multi-image reasoning

Molmo 2 expands the open-model options for video understanding tasks, especially where tracking across frames and reasoning across multiple images matter. For streaming and video tooling teams, that points to a growing AI stack for analyzing visual content rather than just tagging it. The article does not give benchmark scores, pricing, or access limits, so the key signal is whether Ai2 publishes evaluations showing how Molmo 2 performs on video tracking and multi-image reasoning compared with prior open models.