返回首页
AI on Medium··行业媒体

GLM-5.3 Crushed Gemini 3.7 Flash 28.3 to 14.9 After Tying It on DeepSWE

中文摘要

GLM-5.3 在 DeepSWE 测试中以 28.3 分远超 Gemini 3.7 Flash 的 14.9 分,此前两者在其他基准测试中表现持平。

English Summary

GLM-5.3 crushed Gemini 3.7 Flash on the DeepSWE benchmark with 28.3 points versus 14.9, despite tying on two other shared benchmarks.

原文节选

Two coding models launched 24 hours apart. On two shared benchmarks they are within 2.4 points. On the third it is 28.3 to 14.9. Continue reading on Medium »