| agents-on-posttrainbench-v1-1 | Claude Fable 5 (1M, Max) (Claude Code) | #2 | Accuracy: 41.79 |
| agents-on-posttrainbench-v1-1 | GPT-5.6 Sol (Max) (Codex CLI) | #4 | Accuracy: 36.23 |
| agents-on-posttrainbench-v1-1 | Claude Opus 5 (Claude Code) | #6 | Accuracy: 35.04 |
| agents-on-posttrainbench-v1-1 | Kimi K3 (Claude Code) | #7 | Accuracy: 31.96 |
| agents-on-posttrainbench-v1-1 | GLM-5.2 (Max) (Claude Code) | #8 | Accuracy: 31.70 |
| agents-on-posttrainbench-v1-1 | Claude Opus 4.7 (xHigh) (Claude Code) | – | Accuracy: 28.56 |
| agents-on-posttrainbench-v1-1 | Claude Opus 4.8 (High) (Claude Code) | – | Accuracy: 33.84 |
| agents-on-posttrainbench-v1-1 | GPT-5.4 (High) (Codex CLI) | – | Accuracy: 19 |
| agents-on-posttrainbench-v1-1 | GPT-5.5 (xHigh) (Codex CLI) | – | Accuracy: 27.23 |
| agents-on-posttrainbench-v1-1 | Gemini 3.1 Pro (OpenCode) | – | Accuracy: 21.99 |
| agents-on-posttrainbench-v1-1 | Grok 4.5 (High) (Cursor CLI) | – | Accuracy: 23.45 |