In plain words: A 1-trillion-parameter model learned from automatically checked answers alone, with stabilizing tweaks that keep its reasoning readable and efficient. At this size it learned faster and reached higher performance, and spontaneously began self-checking and organizing its reasoning, making hand-written rules unnecessary.
Abstract
Reinforcement learning with verifiable rewards without human-annotated data, often referred to as zero RL, has emerged as a powerful paradigm for eliciting chain-of-thought reasoning. However, due to computational constraints, existing studies are largely restricted to small models, leaving the training dynamics and emergent capabilities at a large scale unexplored. To meaningfully explore this frontier, we aim to elicit high-quality reasoning behaviors from the model. However, we find that naive scaling often suffers from poor readability, token redundancy, and a lack of adaptive reasoning depth. To address these challenges, we present a stable and efficient training pipeline, incorporating algorithmic and system optimizations such as clipped importance sampling, training-inference ratio correction, and mixed-precision control. Our experiments offer three key findings that validate the "bitter lesson" of scaling: (1) scaling to 1T parameters significantly enhances sample efficiency and performance ceilings; (2) the training process progresses sequentially through an initial discovery phase followed by a sharpening phase; and (3) the model spontaneously develops advanced cognitive behaviors, including anthropomorphism, structured formatting, self-verification, parallel reasoning, and context anxiety, rendering hand-crafted heuristics redundant. Evaluated on seven mathematical benchmarks, Ring-2.5-1T-Zero achieves competitive performance. Additionally, to assess CoT quality beyond final-answer correctness, we propose a structured evaluation framework across three dimensions: comprehensibility, reproducibility, and efficiency, where our model demonstrates clear advantages in producing structured and concise reasoning traces. By sharing our observed emergent phenomena, we hope to provide the community with deeper insights into scaling behaviors, particularly at the 1-trillion scale.
Xinyu Tang, Qianggang Cao, Yurou Liu, Yuliang Zhan, Xiaochong Lan, Yifan Li, Yuchen Yan, Han Peng, Zican Dong, Zhenduo Zhang, Tianshu Wang, Xinyu Kong, et al.
arXiv:2607.12395 · cs.CL · submitted Jul 14, 2026 · updated Jul 16, 2026
abstract · pdf · html
> To evaluate comprehensibility quantitatively, we employ an LLM-as-a-Judge framework
This isn’t the worst idea, but it’s still a bit incestuous. Adding an LLM judge to check for hallucinations creates two new kinds of problems: false positives, where your judge hallucinates an incorrect fact, and false negatives, where the judge lets a hallucination slip by.
> We measure reproducibility through knowledge distillation. By fine-tuning a weaker model on the generated CoT traces, we use the downstream performance gain of the student as a proxy.
And my problem here, as a member of the GPU proletariat, is that this just seems incredibly inefficient. In other words, you’re going to generate a bunch of rollouts from your model then wait for the student to train? I guess if you have the compute to train a trillion params then maybe you don’t care.