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

• URL: http://arxiv.org/abs/2512.02556v1
• Date: Tue, 02 Dec 2025 09:25:14 GMT
• Title: DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models
• Abstract summary: We introduce DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance.<n>We introduce DSA, an efficient attention mechanism that substantially reduces computational complexity.<n>By implementing a robust reinforcement learning protocol and scaling post-training compute, DeepSeek-V3.2 performs comparably to GPT-5.
• Score: 219.58681099795186
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• 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.

Related papers

• Video-DeepResearch: Towards the Next-Generation Multimodal Deepresearch Agent [81.19460113054048]
  Video-DR features a decoupled perception-temporal pipeline with stage-wise tool unlocking.<n>Our framework adopts a two-stage training recipe: supervised fine-tuning followed by Group Relative Policy Optimization.<n> Empirical results demonstrate that our Video-ResearchDeepB-A3B establishes a new state-of-the-art of 64.0% average accuracy.
  arXiv  Detail & Related papers  (2026-08-04T17:45:16Z)
• E3-TIR: Enhanced Experience Exploitation for Tool-Integrated Reasoning [26.716382486149453]
  We propose E3-TIR (Enhanced Experience Exploitation), a warm-up paradigm for the early stages of agent training.<n> Specifically, we formulate training as the dynamic integration of three experience types: Expert Prefixes, Expert Guided, and Self-Exploration.<n>Our method dynamically adapts the model's knowledge boundaries, effectively balancing exploration diversity with training efficiency.
  arXiv  Detail & Related papers  (2026-04-10T16:14:48Z)
• AgentCPM-Explore: Realizing Long-Horizon Deep Exploration for Edge-Scale Agents [75.67445299298949]
  AgentCPM-Explore is a compact 4B agent model with high knowledge density and strong exploration capability.<n>We introduce a holistic training framework featuring parameter-space model fusion, reward signal denoising, and contextual information refinement.<n>AgentCPM-Explore achieves state-of-the-art (SOTA) performance among 4B-class models, matches or surpasses 8B-class SOTA models on four benchmarks, and even outperforms larger-scale models such as Claude-4.5-Sonnet or DeepSeek-v3.2 in five benchmarks.
  arXiv  Detail & Related papers  (2026-02-06T08:24:59Z)
• POP: Prefill-Only Pruning for Efficient Large Model Inference [5.743318651374061]
  Large Language Models (LLMs) and Vision-Language Models (VLMs) have demonstrated remarkable capabilities.<n>Existing structured pruning methods, while hardware-efficient, often suffer from significant accuracy degradation.<n>We argue that this failure stems from a stage-agnostic pruning approach that overlooks the asymmetric roles between the prefill and decode stages.
  arXiv  Detail & Related papers  (2026-02-03T09:22:26Z)
• ASTER: Agentic Scaling with Tool-integrated Extended Reasoning [27.877412657068806]
  Reinforcement learning (RL) has emerged as a dominant paradigm for eliciting long-horizon reasoning in Large Language Models (LLMs)<n>We introduce ASTER (Agentic Scaling with Tool-integrated Extended Reasoning), a framework that circumvents this collapse through a targeted cold-start strategy.<n>We find that a small expert cold-start set of just 4K interaction-dense trajectories yields the strongest downstream performance.
  arXiv  Detail & Related papers  (2026-02-01T12:46:02Z)
• How to Set the Learning Rate for Large-Scale Pre-training? [73.03133634525635]
  We formalize this investigation into two distinct research paradigms: Fitting and Transfer.<n>Within the Fitting Paradigm, we introduce a Scaling Law for search factor, effectively reducing the search complexity from O(n3) to O(n*C_D*C_) via predictive modeling.<n>We extend the principles of $$Transfer to the Mixture of Experts (MoE) architecture, broadening its applicability to encompass model depth, weight decay, and token horizons.
  arXiv  Detail & Related papers  (2026-01-08T15:55:13Z)
• Teaching Language Models to Reason with Tools [73.21700643314917]
  We present emphHint-Engineering, a new data synthesis strategy that strategically injects diverse hints at optimal points within reasoning paths.<n>CoRT significantly enhances efficiency, reducing token usage by approximately 30% for the 32B model and 50% for the 1.5B model.
  arXiv  Detail & Related papers  (2025-10-23T08:41:44Z)
• Apriel-1.5-15b-Thinker [19.19917266898226]
  Apriel-1.5-15B-Thinker is a 15-billion parameter open-weights multimodal reasoning model.<n>It achieves frontier-level performance through training design rather than sheer scale.
  arXiv  Detail & Related papers  (2025-10-01T17:29:35Z)
• Iterative Augmentation with Summarization Refinement (IASR) Evaluation for Unstructured Survey data Modeling and Analysis [0.43988112145759295]
  This work introduces a principled evaluation framework for large language model (LLM) based text augmentation.<n> Empirical evaluations show that GPT-3.5 Turbo achieved the best balance of semantic fidelity, diversity, and generation efficiency.
  arXiv  Detail & Related papers  (2025-07-16T10:49:30Z)
• Enhancing Visual Grounding for GUI Agents via Self-Evolutionary Reinforcement Learning [28.291759852111586]
  We introduce a reinforcement learning (RL) based framework that incorporates three core strategies to improve GUI agent performance.<n>With only 3k training samples, our 7B- parameter model achieves state-of-the-art results among similarly sized models.<n> Notably, it attains 47.3% accuracy on the ScreenSpot-Pro dataset, outperforming much larger models, such as UI-TARS-72B, by a margin of 24.2%.
  arXiv  Detail & Related papers  (2025-05-18T11:22:04Z)
• G3Flow: Generative 3D Semantic Flow for Pose-aware and Generalizable Object Manipulation [65.86819811007157]
  We present G3Flow, a novel framework that constructs real-time semantic flow, a dynamic, object-centric 3D representation by leveraging foundation models.<n>Our approach uniquely combines 3D generative models for digital twin creation, vision foundation models for semantic feature extraction, and robust pose tracking for continuous semantic flow updates.<n>Our results demonstrate the effectiveness of G3Flow in enhancing real-time dynamic semantic feature understanding for robotic manipulation policies.
  arXiv  Detail & Related papers  (2024-11-27T14:17:43Z)
• ALOcc: Adaptive Lifting-Based 3D Semantic Occupancy and Cost Volume-Based Flow Predictions [91.55655961014027]
  3D semantic occupancy and flow prediction are fundamental to understanding scene scene.<n>This paper proposes a vision-based framework with three targeted improvements.<n>Our purely convolutional architecture establishes new SOTA performance on multiple benchmarks for both semantic occupancy and joint semantic-flow prediction.
  arXiv  Detail & Related papers  (2024-11-12T11:32:56Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.