Soham Dan, Jeff Da, Sanyam Bhutani, Miguel Romero Calvo, Iván Guevara, Fernando Crabedo, José Bernal, Julio Reyna, Luis Leandro Monsalve Mendez, Ying Liu, Daniel Yue Zhang · 2026 · arXiv (Cornell University)
DOI: 10.48550/arxiv.2610.24206 · cited by 0 works (OpenAlex) · open-access PDF
SWE-Bench Pro est un benchmark largement utilisé des tâches d'ingénierie logicielle à longue portée pour les agents de codage. À mesure que les agents s'améliorent, ils atteignent les problèmes de tâches que les agents antérieurs n'ont jamais touchés et interagissent avec l'environnement d'évaluation de la manière dont l'appareil d'origine n'avait pas prévu. Nous présentons SWE-Bench Pro V2, une division publique révisée construite pour garder le benchmark précis à mesure que les capacités avancent. Les audits d'agent, l'adjudication d'experts et la réimplémentation aveugle sont réparés ou retirés 480 des 731 tâches: 23 ingénieurs contractés ont passé 404 heures, travaillant chacun dans une lang
Machine translation (Meta M2M100); the English original follows.
SWE-Bench Pro is a widely used benchmark of long-horizon software engineering tasks for coding agents. As agents improve, they reach task issues earlier agents never touched and interact with the evaluation environment in ways the original harness did not anticipate. We present SWE-Bench Pro V2, a revised public split built to keep the benchmark accurate as capabilities advance. Agentic audits, expert adjudication and blind re-implementation repaired or retired 480 of the 731 tasks: 23 contracted engineers spent 404 hours, each working in a language they write professionally, and every revised instruction was verified by a second engineer solving the task from the text alone. A hardened protocol restricts network access to the model endpoint and re-grades every patch in a clean container, so agents cannot retrieve upstream fixes or get credit for changes outside their patch. On the 642 V2 tasks frontier agents resolve up to 99.4% while other systems stay clearly separated, and a 51-task Hard split spreads systems from 52.9% to 98.0%. We release the benchmark and distill lessons for maintaining agent benchmarks over time.
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
Topics: Benchmark (surveying), Computer science, Coding (social sciences), Task (project management), Software, Software engineering
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.24206, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange