A System One Model for Fast and Generalizable Decision-Making

Jacky Kwok$^{\dagger}$, Hangoo Kang, Tarun Suresh, Jon Saad-Falcon, Marco Pavone

Christopher Ré, Azalia Mirhoseini :stanford:Stanford University :nvidia:NVIDIA Research

$^{\dagger}$ Project Lead

🗓️ Posted: Sep 23, 2026

<aside> ⚡

New Architecture, Data Recipe, and Scaling Laws

• We introduce Contrastive Language Models (CLMs), a new class of System One model trained with a contrastive learning objective that connects states and actions.
• We release CLM-8B, which is pre-trained on 60M Nemotron Q&A pairs, mid-trained on 30M synthetic hard negatives, and post-trained on 1M agentic trajectories.
• CLM-8B delivers performance comparable to Jev across computer-use, gaming, and tool-calling tasks, while achieving up to 9× lower latency. When fine-tuned as a trajectory reward model, CLM also sets a new SOTA on challenging agentic coding benchmarks, including DeepSWE (81.6%) and Terminal Bench 2.1 (87.6%).
• We build an ultra-efficient training and serving infra for CLM by disaggregating states and actions, allowing their embeddings to be cached and reused independently.
• We establish scaling laws for CLMs and show that the test contrastive loss decreases predictably as a power law in training compute, model size, and dataset size.

Try CLM on GitHub:

</aside>

Overview

CLM first trains a state encoder and an action encoder on a large-scale dataset with a contrastive objective (InfoNCE), so that each state is pulled toward the ground truth action that was taken and pushed away from all others. The two encoders then serve directly as a zero-shot action classifier.

At deployment, given the current state and a set of candidate actions, CLM scores each action by how well its embedding aligns with the state embedding and selects the highest-scoring action.

Dino Run (CLM vs. Jev)

• Super Mario Demo
• WikiRacing Demo

Zero-shot Evaluation

Across computer-use, gaming, and tool-calling tasks, CLM-8B performs on par with Jev while running up to 9× faster. The speedups are most pronounced when the number of candidate actions is large (e.g., WikiRacing) or when actions can be frequently reused across states (e.g., T-Rex Game).