AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Evaluating Spatiotemporal Reasoning of Vision-Language Models in Atari Gameplay

Mingjia Huo, Yao Fu, Bo Chang, Yaqing Wang, Dawei Zhu, Dylan Zhang, Xinyang Yi, Lichan Hong, Ed Chi, Hao Zhang, Jiaxi Tang

NeurIPS 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Reinforcement learning for visual reasoning

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Introduces AtariBench, a controlled benchmark evaluating vision-language models on sequential decision-making across 20 Atari 2600 games from high-temporal-resolution gameplay streams.

Problem

VLMs perform well on static visual understanding benchmarks, but their ability to act in dynamic, fast-changing environments is poorly understood.

Approach

A controlled Atari 2600 benchmark that decouples fast in-game dynamics from wall-clock time so VLMs are evaluated on sequential decisions from high-temporal-resolution gameplay streams.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use append-only full history instead of structured history (ablation)Gemini 2.5 Flash (off) · Breakoutnoop-only share of turns: 0.7% -> 48.8%Hurt
use append-only history (ablation)Gemini 2.5 Flash/Pro, Gemini 3 Flash · Air Raid, Pacman, PhoenixrewardNo clear effect
add textual reward and life-loss feedback (ablation)Gemini 2.5 Flash · Breakoutreward: 1.5 -> 2.3Improved
add textual feedback (ablation)Gemini 2.5 Flash · Fishing Derbyreward: -11.8 -> -9.0Improved
add textual feedback (ablation)Gemini 3.1 Pro · four gamesrewardNo clear effect
allow 1-10 actions per query instead of 1 action (ablation)Gemini 3 Flash · Breakoutreward: 4.6 -> 3.7Hurt
vary action-chunk length (ablation)Gemini 3 Flash and GPT-5.4 Mini · three other game pairs and modelsrewardNo clear effect
raise reasoning effortVLMs · AtariBenchscore and diagnostic accuracyNo clear effect
re-query with current frame only instead of complete history (ablation)VLM on spatial-relation failure cases · Freeway etc.spatial error persistence: 39.3% -> 10.7%Improved
use fixed 30-second horizon instead of game-over termination (ablation)Gemini 2.5 Flash · DemonAttackper-run reward SD: 36.3 -> 13.7Improved
add survival instruction to prompt under game-over termination (ablation)Gemini 2.5 Flash · Phoenixaverage reward: 177.9 -> 195.4Improved

Experimental setup

Datasets: Atari 2600 gameplay (20 games)

Benchmarks: AtariBench

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy