| Benchmark | Model | Rank | Results |
|---|---|---|---|
| agents-on-bench-telecom | KAT-Coder-V2 | – | Accuracy: 93.9 |
| agents-on-pinchbench | KAT-Coder-V2 | – | Accuracy: 88.7 |
| code-generation-on-swe-bench-multilingual | KAT-Coder-V2 (Claude Code) | – | Accuracy: 75.4 |
| code-generation-on-swe-bench-verified | KAT-Coder-V2 (Claude Code) | – | Accuracy: 79.6 |
| instruction-following-on-ifbench | KAT-Coder-V2 | – | Accuracy: 67.0 |