| Benchmark | Model | Rank | Results |
|---|---|---|---|
| computer-use-agents-on-osworld-2-0 | Qwen 3.7-Plus + LongHorizon-Harness | #7 | Success Rate: 8.3 |
| code-generation-on-terminal-bench-2-1 | GPT-5.6 Luna + LongHorizon-Harness (LongHorizon-Harness (Codex executor)) | #12 | Accuracy: 83.1 |