AM-Thinking-v1: Advancing the Frontier of Reasoning at 32B Scale

2025-05-13

AI Summary

To address the insufficient mathematical and code reasoning capabilities of medium-scale (32B) open-source language models, this work introduces AM-Thinking-v1—the first dense-architecture 32B model achieving performance on par with hundred-billion-parameter Mixture-of-Experts (MoE) models in mathematical and coding reasoning. Built upon the Qwen2.5-32B base architecture, AM-Thinking-v1 employs a novel synergistic post-training paradigm integrating high-quality supervised fine-tuning (SFT) and reinforcement learning (RL), leveraging publicly available reasoning datasets. It achieves state-of-the-art scores among open-source models of comparable scale: 85.3 on AIME 2024, 74.4 on AIME 2025, and 70.3 on LiveCodeBench. These results demonstrate that carefully optimized dense models at the 32B scale can effectively balance reasoning capability, deployment efficiency, and open collaboration—without requiring MoE sparsity or excessive parameter counts. The model, training recipes, and evaluation protocols are fully open-sourced.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Large Vision Models
• Natural Language Processing: (Large) Language Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

We present AM-Thinking-v1, a 32B dense language model that advances the frontier of reasoning, embodying the collaborative spirit of open-source innovation. Outperforming DeepSeek-R1 and rivaling leading Mixture-of-Experts (MoE) models like Qwen3-235B-A22B and Seed1.5-Thinking, AM-Thinking-v1 achieves impressive scores of 85.3 on AIME 2024, 74.4 on AIME 2025, and 70.3 on LiveCodeBench, showcasing state-of-the-art mathematical and coding capabilities among open-source models of similar scale. Built entirely from the open-source Qwen2.5-32B base model and publicly available queries, AM-Thinking-v1 leverages a meticulously crafted post-training pipeline - combining supervised fine-tuning and reinforcement learning - to deliver exceptional reasoning capabilities. This work demonstrates that the open-source community can achieve high performance at the 32B scale, a practical sweet spot for deployment and fine-tuning. By striking a balance between top-tier performance and real-world usability, we hope AM-Thinking-v1 inspires further collaborative efforts to harness mid-scale models, pushing reasoning boundaries while keeping accessibility at the core of innovation. We have open-sourced our model on href{https://huggingface.co/a-m-team/AM-Thinking-v1}{Hugging Face}.

Problem

• Advancing reasoning capabilities in 32B-scale language models
• Achieving state-of-the-art mathematical and coding performance
• Demonstrating open-source potential for mid-scale model deployment

Innovation

• 32B dense language model for advanced reasoning
• Post-training pipeline with fine-tuning and reinforcement learning
• Open-source Qwen2.5-32B base model utilization