Responsive captures
360 / 768 / 1440
Z.ai / Zhipu AI
GLM-5.2
Frozen one-attempt benchmark evidence. The live build is reproduced from the untouched model worktree using deployment-path adaptations recorded in its manifest.
Total49/100
Raw91.75/100
Automated56/60
Visual17/20
Code8.75/10
GateFail
Interaction evidence
Run facts
- Status
- completed
- Duration
- 6.8 minutes
- Cost
- $0.1486 USD
- Agent steps
- 21
- Candidate
- C-6833
Failed automated checks
- F05[F05|4][GATE] the native-select form has errors and deterministic successhard gate