DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models

TL;DR

DeepSeek-V3.2 advances open large language models by introducing efficient attention, scalable reinforcement learning, and a large-scale data synthesis pipeline, achieving state-of-the-art reasoning and agent performance.

Contribution

It presents novel attention mechanisms, a scalable reinforcement learning framework, and a large-scale data synthesis pipeline, significantly improving efficiency and reasoning in large language models.

Findings

Surpasses GPT-5 in performance

Achieves IMO and IOI gold medals

Enhances reasoning and instruction-following robustness

Abstract

We introduce DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance. The key technical breakthroughs of DeepSeek-V3.2 are as follows: (1) DeepSeek Sparse Attention (DSA): We introduce DSA, an efficient attention mechanism that substantially reduces computational complexity while preserving model performance in long-context scenarios. (2) Scalable Reinforcement Learning Framework: By implementing a robust reinforcement learning protocol and scaling post-training compute, DeepSeek-V3.2 performs comparably to GPT-5. Notably, our high-compute variant, DeepSeek-V3.2-Speciale, surpasses GPT-5 and exhibits reasoning proficiency on par with Gemini-3.0-Pro, achieving gold-medal performance in both the 2025 International Mathematical Olympiad (IMO) and the International Olympiad in Informatics (IOI). (3) Large-Scale Agentic Task Synthesis Pipeline: To integrate reasoning into tool-use scenarios, we developed a novel synthesis pipeline that systematically generates training data at scale. This methodology facilitates scalable agentic post-training, yielding substantial improvements in generalization and instruction-following robustness within complex, interactive environments.