The Year in AI Papers
The year/Independent research

Paper 2511.18538

From Code Foundation Models to Agents and Applications: A Comprehensive Survey and Practical Guide to Code Intelligence

Jian Yang, Xianglong Liu, Weifeng Lv, Ken Deng, Shawn Guo, Lin Jing, Yizhi Li, Shark Liu, Xianzhen Luo, Yuyu Luo, Changzai Pan, Ensheng Shi, et al.

Published
Nov 2025
Research lab
Independent
Citations
12
GitHub
Not linked

01 In brief

Summary

This survey comprehensively examines the lifecycle of code-focused large language models (LLMs), from data curation and pre-training to post-training, alignment, and deployment as autonomous agents.

It analyzes both general-purpose LLMs (e.g., GPT-4, Claude, LLaMA) and code-specialized models (e.g., StarCoder, Code LLaMA, DeepSeek-Coder, QwenCoder), covering model architectures, training stages, and open-source datasets.

The paper reviews a wide range of code tasks, benchmarks, and evaluation metrics, including function-level and repository-level tasks, and discusses alignment techniques such as supervised fine-tuning (SFT) and reinforcement learning (RL), particularly with verifiable rewards (RLVR).

It explores software engineering agents across the development lifecycle, the role of code in generalist agents, and safety considerations for code LLMs.

The survey also provides practical training recipes, including distributed training frameworks and hyperparameter guidelines, and concludes with an overview of applications like IDE-integrated assistants and cloud-native platforms.

The work bridges the gap between academic research and real-world deployment, offering a strategic roadmap for researchers and practitioners.

02 From the paper

Abstract

Large language models (LLMs) have fundamentally transformed automated software development by enabling direct translation of natural language descriptions into functional code, driving commercial adoption through tools like Github Copilot (Microsoft), Cursor (Anysphere), Trae (ByteDance), and Claude Code (Anthropic). While the field has evolved dramatically from rule-based systems to Transformer-based architectures, achieving performance improvements from single-digit to over 95\% success rates on benchmarks like HumanEval. In this work, we provide a comprehensive synthesis and practical guide (a series of analytic and probing experiments) about code LLMs, systematically examining the complete model life cycle from data curation to post-training through advanced prompting paradigms, code pre-training, supervised fine-tuning, reinforcement learning, and autonomous coding agents. We analyze the code capability of the general LLMs (GPT-4, Claude, LLaMA) and code-specialized LLMs (StarCoder, Code LLaMA, DeepSeek-Coder, and QwenCoder), critically examining the techniques, design decisions, and trade-offs. Further, we articulate the research-practice gap between academic research (e.g., benchmarks and tasks) and real-world deployment (e.g., software-related code tasks), including code correctness, security, contextual awareness of large codebases, and integration with development workflows, and map promising research directions to practical needs. Last, we conduct a series of experiments to provide a comprehensive analysis of code pre-training, supervised fine-tuning, and reinforcement learning, covering scaling law, framework selection, hyperparameter sensitivity, model architectures, and dataset comparisons.