Verified65% confidenceBenchmarkExact time
SWE-bench requires models to resolve real software defects on GitHub and is the authoritative benchmark for evaluating agent-level programming capabilities.
3
Sources
65%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
Related Claims
UnverifiedSWE-bench extracts tasks from real GitHub Issues, requiring models to fix real bugs in mainstream open-source projects, and is regarded as the gold standard for measuring the ability to actually solve engineering problems.82% similarUnverifiedSWE-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要求模型在真实GitHub仓库中定位问题并生成可通过CI测试的补丁，以Resolve Rate作为核心指标81% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/564993API
curl https://kongchang.com/api/v1/knowledge/claims/564993MCP
get_claim(id=564993)