The Year in AI Papers
The year/Google DeepMind

Paper 2509.20328

Video models are zero-shot learners and reasoners

Thaddäus Wiedemer, Yuxuan Li, Paul Vicol, Shixiang Shane Gu, Nick Matarese, Kevin Swersky, Been Kim, Priyank Jaini, Robert Geirhos

Published
Sep 2025
Research lab
Google DeepMind
Citations
194
GitHub
Not linked

01 In brief

Summary

This paper investigates whether generative video models, like Veo 3, can act as zero-shot learners and reasoners for general-purpose vision tasks, similar to how LLMs transformed NLP.

The authors analyzed 18,384 generated videos across 62 qualitative and 7 quantitative tasks, finding that Veo 3 can solve tasks it wasn't explicitly trained for, including edge detection, segmentation, image editing, physical property understanding, object affordances, and tool use.

These abilities enable early forms of visual reasoning, such as maze and symmetry solving, through a 'chain-of-frames' approach.

Quantitative results show Veo 3 outperforms its predecessor Veo 2 significantly, often matching or exceeding the image editing model Nano Banana, though it still lags behind task-specific models like SAMv2.

The paper argues that video models are on a trajectory to become unified vision foundation models, with performance improvements and inference-time scaling suggesting rapid progress.

However, the authors note that performance is a lower bound due to prompt sensitivity and that video generation costs, while currently high, are expected to fall.

02 From the paper

Abstract

The remarkable zero-shot capabilities of Large Language Models (LLMs) have propelled natural language processing from task-specific models to unified, generalist foundation models. This transformation emerged from simple primitives: large, generative models trained on web-scale data. Curiously, the same primitives apply to today's generative video models. Could video models be on a trajectory towards general-purpose vision understanding, much like LLMs developed general-purpose language understanding? We demonstrate that Veo 3 can solve a broad variety of tasks it wasn't explicitly trained for: segmenting objects, detecting edges, editing images, understanding physical properties, recognizing object affordances, simulating tool use, and more. These abilities to perceive, model, and manipulate the visual world enable early forms of visual reasoning like maze and symmetry solving. Veo's emergent zero-shot capabilities indicate that video models are on a path to becoming unified, generalist vision foundation models.