Unverified85% confidenceFactTime unknown
SWE-Bench Verified是经人工验证的500道高质量子集，SWE-Bench Lite是300道较简单的子集
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedQwen3.6-27B, with 27B parameters, surpassed the performance of the previous-generation 397B MoE flagship model on multiple benchmarks including SWE Bench Verified, SWE Bench Pro, and Terminal Bench 2.0.61% similarUnverifiedThe top scorer on the SWE-Bench engineering practice test is the recently released MiniMax M2 (397B).60% similarUnverifiedStep 3.7 Flash 在 SWE-Bench PRO 基准测试中以 56.3 分排名第二58% similarUnverifiedThe median prompt length for SWE-Bench Pro is 30,098 characters, while the median for the task description portion of Frontier Code is only 9,082 characters.56% similarUnverifiedClaude 3.7 Sonnet achieved 70.3% accuracy on SWE-Bench Verified with custom scaffolding54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/28371API
curl https://kongchang.com/api/v1/knowledge/claims/28371MCP
get_claim(id=28371)