Unverified60% confidenceFactExact time
SWE-bench extracts real GitHub Issues from mainstream Python open-source projects and requires models to generate code patches that can pass official test suites.
2
Sources
60%
Confidence
Long-term
Relevance
7/14/2026
First Seen
Sources
GPT-5.6 Sol XHigh实测：让Claude用户改变立场的编程能力
Related Claims
VerifiedSWE-bench was created by a research team at Princeton University, drawing issues from real open-source GitHub projects and requiring models to generate patches that pass test cases.82% similarVerifiedSWE-bench directly extracts issues and corresponding pull requests from real GitHub open-source projects, requiring AI systems to understand problem descriptions, locate code defects, and generate correct fix patches80% similarVerifiedSWE-bench requires models to autonomously complete code modifications starting from real GitHub issues and pass test cases, making it closer to real-world engineering scenarios than HumanEval.79% similarVerifiedSWE-bench Verified selects bug reports with clear solutions from GitHub Issues of real open-source projects, where the "Verified" suffix indicates that the problems have been manually validated.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/503437API
curl https://kongchang.com/api/v1/knowledge/claims/503437MCP
get_claim(id=503437)