| Benchmark | Model | Rank | Results |
|---|---|---|---|
| mathematical-reasoning-on-aime-2026 | PACT (Qwen3.5-4B) | #30 | Accuracy: 85.21 |
| mathematical-reasoning-on-aime-2025 | PACT (Qwen3.5-4B) | #35 | Accuracy: 83.12 |
| code-generation-on-swe-bench-verified | PACT (Qwen3.6-35B-A3B) (Codex through Harbor) | #52 | Accuracy: 67.4 |