Saved in:
Bibliographic Details
Main Authors: Golchinfar, David, Vaziri, Daryoush, Marquardt, Alexander
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.07385
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913016548687872
author Golchinfar, David
Vaziri, Daryoush
Marquardt, Alexander
author_facet Golchinfar, David
Vaziri, Daryoush
Marquardt, Alexander
contents We present SauerkrautLM-Doom-MultiVec, a 1.3 million parameter model that plays the classic first-person shooter DOOM in real time, outperforming large language models up to 92,000x its size, including Nemotron-120B, Qwen3.5-27B, and GPT-4o-mini. Our model combines a ModernBERT encoder with hash embeddings, depth-aware token representations, and an attention pooling classification head to select game actions from ASCII frame representations at 31ms per decision. Trained on just 31,000 human gameplay demonstrations, it achieves 178 frags in 10 episodes (17.8 per episode) in the defend_the_center scenario, more than all tested LLMs combined (13 frags total). All agents receive equivalent input: ASCII frames and depth maps. Despite having 92,000x fewer parameters than Nemotron-120B, our model is the only agent that actively engages enemies rather than purely evading them. These results demonstrate that small, task-specific models trained on domain-appropriate data can decisively outperform general-purpose LLMs at real-time control tasks, at a fraction of the inference cost, with deployment capability on consumer hardware.
format Preprint
id arxiv_https___arxiv_org_abs_2604_07385
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Playing DOOM with 1.3M Parameters: Specialized Small Models vs Large Language Models for Real-Time Game Control
Golchinfar, David
Vaziri, Daryoush
Marquardt, Alexander
Machine Learning
Artificial Intelligence
We present SauerkrautLM-Doom-MultiVec, a 1.3 million parameter model that plays the classic first-person shooter DOOM in real time, outperforming large language models up to 92,000x its size, including Nemotron-120B, Qwen3.5-27B, and GPT-4o-mini. Our model combines a ModernBERT encoder with hash embeddings, depth-aware token representations, and an attention pooling classification head to select game actions from ASCII frame representations at 31ms per decision. Trained on just 31,000 human gameplay demonstrations, it achieves 178 frags in 10 episodes (17.8 per episode) in the defend_the_center scenario, more than all tested LLMs combined (13 frags total). All agents receive equivalent input: ASCII frames and depth maps. Despite having 92,000x fewer parameters than Nemotron-120B, our model is the only agent that actively engages enemies rather than purely evading them. These results demonstrate that small, task-specific models trained on domain-appropriate data can decisively outperform general-purpose LLMs at real-time control tasks, at a fraction of the inference cost, with deployment capability on consumer hardware.
title Playing DOOM with 1.3M Parameters: Specialized Small Models vs Large Language Models for Real-Time Game Control
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2604.07385