The Limits of Long-Context Reasoning in Automated Bug Fixing

TL;DR

This paper systematically evaluates the long-context reasoning capabilities of current large language models in automated bug fixing, revealing significant limitations despite recent performance gains in software engineering benchmarks.

Contribution

It provides a comprehensive analysis showing that current LLMs struggle with genuine long-context reasoning in bug fixing, emphasizing the gap between nominal and effective context lengths.

Findings

Performance drops sharply at 64k token context

Agentic success mainly from short-context task decomposition

Systematic failure modes include hallucinated diffs and incorrect targets

Abstract

Rapidly increasing context lengths have led to the assumption that large language models (LLMs) can directly reason over entire codebases. Concurrently, recent advances in LLMs have enabled strong performance on software engineering benchmarks, particularly when paired with agentic workflows. In this work, we systematically evaluate whether current LLMs can reliably perform long-context code debugging and patch generation. Using SWE-bench Verified as a controlled experimental setting, we first evaluate state-of-the-art models within an agentic harness (mini-SWE-agent), where performance improves substantially: GPT-5-nano achieves up to a 31\% resolve rate on 100 samples, and open-source models such as Deepseek-R1-0528 obtain competitive results. However, token-level analysis shows that successful agentic trajectories typically remain under 20k-30k tokens, and that longer accumulated contexts correlate with lower success rates, indicating that agentic success primarily arises from task decomposition into short-context steps rather than effective long-context reasoning. To directly test long-context capability, we construct a data pipeline where we artificially inflate the context length of the input by placing the relevant files into the context (ensuring perfect retrieval recall); we then study single-shot patch generation under genuinely long contexts (64k tokens). Despite this setup, performance degrades sharply: Qwen3-Coder-30B-A3B achieves only a 7\% resolve rate at 64k context, while GPT-5-nano solves none of the tasks. Qualitative analysis reveals systematic failure modes, including hallucinated diffs, incorrect file targets, and malformed patch headers. Overall, our findings highlight a significant gap between nominal context length and usable context capacity in current LLMs, and suggest that existing agentic coding benchmarks do not meaningfully evaluate long-context reasoning.