Game-TARS: Pretrained Foundation Models for Scalable Generalist Multimodal Game Agents

TL;DR

Game-TARS introduces a scalable, multimodal pre-trained model for generalist game agents that outperforms previous models and approaches human-level performance across diverse gaming environments.

Contribution

It presents a unified, scalable action space and pre-training methodology enabling large-scale continual learning across heterogeneous game domains.

Findings

Achieves twice the success rate on Minecraft tasks.

Performs comparably to humans in unseen web 3D games.

Outperforms GPT-5, Gemini-2.5-Pro, and Claude-4-Sonnet in FPS benchmarks.

Abstract

We present Game-TARS, a generalist game agent trained with a unified, scalable action space anchored to human-aligned native keyboard-mouse inputs. Unlike API- or GUI-based approaches, this paradigm enables large-scale continual pre-training across heterogeneous domains, including OS, web, and simulation games. Game-TARS is pre-trained on over 500B tokens with diverse trajectories and multimodal data. Key techniques include a decaying continual loss to reduce causal confusion and an efficient Sparse-Thinking strategy that balances reasoning depth and inference cost. Experiments show that Game-TARS achieves about 2 times the success rate over the previous sota model on open-world Minecraft tasks, is close to the generality of fresh humans in unseen web 3d games, and outperforms GPT-5, Gemini-2.5-Pro, and Claude-4-Sonnet in FPS benchmarks. Scaling results on training-time and test-time confirm that the unified action space sustains improvements when scaled to cross-game and multimodal data. Our results demonstrate that simple, scalable action representations combined with large-scale pre-training provide a promising path toward generalist agents with broad computer-use abilities.