# Thinker: A vision-language foundation model for embodied intelligence

Source: [Thinker: A vision-language foundation model for embodied intelligence](<https://www.sota2.com/research/paper/697c3af0a8b01e76cc8da39a>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-31T05:00:04.715Z

Authors: Baiyu Pan, Daqin Luo, Junpeng Yang, Jiyuan Wang, Yixuan Zhang, Hailin Shi, Jichao Jiao

Published: 2026-01-29T02:52:08.000Z

## Abstract

When large vision-language models are applied to the field of robotics, they encounter problems that are simple for humans yet error-prone for models. Such issues include confusion between third-person and first-person perspectives and a tendency to overlook information in video endings during temporal reasoning. To address these challenges, we propose Thinker, a large vision-language foundation model designed for embodied intelligence. We tackle the aforementioned issues from two perspectives. Firstly, we construct a large-scale dataset tailored for robotic perception and reasoning, encompassing ego-view videos, visual grounding, spatial understanding, and chain-of-thought data. Secondly, we introduce a simple yet effective approach that substantially enhances the model's capacity for video comprehension by jointly incorporating key frames and full video sequences as inputs. Our model achieves state-of-the-art results on two of the most commonly used benchmark datasets in the field of task planning.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Robotic Video Question Answering on RoboVQA (test)](<https://www.sota2.com/research/sota/robotic-video-question-answering-on-robovqa-test.md>): Robotic Video Question Answering; RoboVQA (test); BLEU-1: 72.7.

- [Egocentric Action Planning on EgoPlan-bench v2 (test)](<https://www.sota2.com/research/sota/egocentric-action-planning-on-egoplan-bench-v2-test.md>): Egocentric Action Planning; EgoPlan-bench v2 (test); Daily life Success Rate: 63.78.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2601.21199>)

[PDF](<https://arxiv.org/pdf/2601.21199.pdf>)