Verified65% confidenceFactExact time
SWE-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.
3
Sources
65%
Confidence
Long-term
Relevance
7/12/2026
First Seen
Sources
Alibaba Bans Claude Code: The Battle Over AI Coding Tool Security and Data Sovereignty
bilibiliSynthMindX7/4/2026
Related Claims
UnverifiedSWE-bench评估模型解决真实GitHub Issue的能力，被认为比HumanEval更贴近Agent实际工作场景82% similarUnverifiedSWE-bench requires models to read issues from real GitHub repositories, autonomously generate patches, and pass validation on a hidden test set, and is considered one of the benchmarks most closely reflecting real-world engineering capability.80% similarVerifiedSWE-bench requires models to resolve real software defects on GitHub and is the authoritative benchmark for evaluating agent-level programming capabilities.80% similarUnverifiedSWE-bench extracts real GitHub Issues from mainstream Python open-source projects and requires models to generate code patches that can pass official test suites.79% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/489178API
curl https://kongchang.com/api/v1/knowledge/claims/489178MCP
get_claim(id=489178)