Benchmark
PostTrainBench
Benchmark by PostTrainBench team
PostTrainBench v1.1 · September 2026 results
31.70%
GLM 5.2, Max reasoning / Claude Code
frontier lab
Z.ai
Organization attribution for GLM 5.2; benchmark agent used Claude Code.
Published evidence
Benchmark
PostTrainBench
Benchmark by PostTrainBench team
PostTrainBench v1.2 · September 30 release
36.79%
GLM 5.3 Flash / Claude Code, Max
Benchmark
PostTrainBench
Benchmark by PostTrainBench team
PostTrainBench v1.2 · September 30 release
34.52%
GLM 5.3 / Claude Code, Max
Benchmark
PostTrainBench
Benchmark by PostTrainBench team
PostTrainBench v1.2 · September 30 release
30.40%
GLM 5.2, Max reasoning / Claude Code
Benchmark
TasteVal
Benchmark by P-Zero Research
October 2026 paper v1 · 20-model main cohort
0.59× ×
GLM 5.3 (max) / TasteVal
Benchmark
TasteVal
Benchmark by P-Zero Research
October 2026 paper v1 · 20-model main cohort
0.89× ×
GLM 5.3 (max) / TasteVal