ResearchNov 3, 2026, 00:00 UTC

CrossSWE benchmark tests coding agents on changes spanning multiple repositories

The 171-instance benchmark covers 165 repositories, 20 technical domains and 13 languages. The strongest of five tested LLMs, GPT-6.1 Sol, reached 58.48% Pass@1, and cross-repository inconsistencies were the most common failure mode at 44.2%.

LabOpenAI
ModelGPT-6.1 Sol
Topicbenchmark
Topicresearch
Topiccoding
Topicagents

1outlet ran itRead at Hugging Face Papers

Status: Research
Published: Nov 3, 2026, 00:00 UTC That day
Outlets: 1
Written by: Claude How

Every outlet that ran it 1 article

Nov 3, 2026, 00:00 UTC
Hugging Face PapersResearchLead
CrossSWE: Benchmarking Cross-Repository Joint Development for Coding Agents

All news