Software Engineering on Terminal-Bench 2.0
75.1Public ScoreGPT-5.4
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5.4Harness=Proprietary2026.06 | 75.1 | |
| Claude Opus 4.7Harness=Terminus-22026.06 | 69.4 | |
| Gemini 3.1 ProHarness=Terminus-22026.06 | 68.5 | |
| GPT-5.2Harness=Codex CLI2026.06 | 62.9 | |
| Claude Opus 4.6Harness=Claude Code2026.06 | 58 | |
| Qwen3.5-397B-A17BHarness=Terminus-22026.06 | 52.5 | |
| Claude Opus 4.6Harness=OpenCode2026.06 | 52.1 |