Coding on CyberGym
50.6ScoreClaude Opus 4.5
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude Opus 4.5Framework=Claude Code 2.1.182026.02 | 50.6 | |
| GLM-5Framework=Claude Code 2.1.182026.02 | 43.2 | |
| Kimi K2.5Framework=Claude Code 2.1.182026.02 | 41.3 | |
| Gemini 3 ProFramework=Claude Code 2.1.182026.02 | 39.9 | |
| GLM-4.7Framework=Claude Code 2.1.182026.02 | 23.5 | |
| DeepSeek-V3.2Framework=Claude Code 2.1.182026.02 | 17.3 |