Home
Shanghai Jiao Tong University Team Releases SWE-Explore Benchmark, Exposing Line-Level Localization Flaws in AI Coding Agents
An international research team, including Shanghai Jiao Tong University, today launched SWE-Explore, a new benchmark testing tool. It quantitatively reveals significant technical shortcomings of current AI coding agents at the "line-level accuracy" by decoupling code search from the actual repair phase. This study moves beyond the previous single evaluation model that relied solely on the "final repair rate," offering a new standard for directly measuring the quality of upstream search in agents, and driving the evolution of AI software engineering evaluation toward deeper areas.
Traditional benchmarks like SWE-bench often mask agents' real defects in code reading and understanding stages because they focus only on end-to-end results. To address this, the research team extracted consensus code segments from multiple independent solution paths based on the successful operation trajectories of mainstream large models such as GPT-5.4, Gemini3Pro, Claude Sonnet4.6, and Kimi K2.6, building a dataset containing 848 defect tasks across 10 programming languages and 203 open-source projects.

Evaluation results show that although general coding agents like Claude Code and OpenHands perform well in "file-level" positioning, their core area coverage drops sharply to between 14% and 19% when focusing on specific "code lines." Ablation experiments further confirmed the "minimum context threshold" effect: when the visible proportion of key core areas is below 50%, the model's repair generally fails; however, once it crosses the threshold of 50% to 75%, the repair success rate increases dramatically.
This research result indicates that the current bottleneck for AI agents is not entirely about patch writing capability but rather about accurately filtering and capturing critical context. In the current industry context, where project managers reject half of automated adoption proposals, the "less filtering, more reading" technical direction proposed by SWE-Explore not only points the way for architecture optimization of next-generation specialized code localization systems (such as CoSIL) but also accelerates the paradigm shift of automated software engineering from "brute-force generation" to "precise retrieval."
Related article
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage
California AV Compliance: A New Era of Tickets, Geofences, and 1M Miles
Guident operates an AuveTech shuttle in South Florida, managing a four-mile route in West Palm Beach and a one-mile route in Boca Raton using its remote monitoring technology. | Credit: GuidentCalifornia is redefining the regulatory landscape for dri
Related Special Topic Recommendations
Comments (0)
0/500
An international research team, including Shanghai Jiao Tong University, today launched SWE-Explore, a new benchmark testing tool. It quantitatively reveals significant technical shortcomings of current AI coding agents at the "line-level accuracy" by decoupling code search from the actual repair phase. This study moves beyond the previous single evaluation model that relied solely on the "final repair rate," offering a new standard for directly measuring the quality of upstream search in agents, and driving the evolution of AI software engineering evaluation toward deeper areas.
Traditional benchmarks like SWE-bench often mask agents' real defects in code reading and understanding stages because they focus only on end-to-end results. To address this, the research team extracted consensus code segments from multiple independent solution paths based on the successful operation trajectories of mainstream large models such as GPT-5.4, Gemini3Pro, Claude Sonnet4.6, and Kimi K2.6, building a dataset containing 848 defect tasks across 10 programming languages and 203 open-source projects.

Evaluation results show that although general coding agents like Claude Code and OpenHands perform well in "file-level" positioning, their core area coverage drops sharply to between 14% and 19% when focusing on specific "code lines." Ablation experiments further confirmed the "minimum context threshold" effect: when the visible proportion of key core areas is below 50%, the model's repair generally fails; however, once it crosses the threshold of 50% to 75%, the repair success rate increases dramatically.
This research result indicates that the current bottleneck for AI agents is not entirely about patch writing capability but rather about accurately filtering and capturing critical context. In the current industry context, where project managers reject half of automated adoption proposals, the "less filtering, more reading" technical direction proposed by SWE-Explore not only points the way for architecture optimization of next-generation specialized code localization systems (such as CoSIL) but also accelerates the paradigm shift of automated software engineering from "brute-force generation" to "precise retrieval."
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage











