| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| CUDA Kernel Generation | KernelBench Level 3 | Executions Count24 | 31 | |
| CUDA Kernel Generation | KernelBench Level 2 | Execution Count48 | 31 | |
| CUDA Kernel Generation | KernelBench Level 1 | Exec Count46 | 31 | |
| Kernel Optimization | KernelBench 1.0 (test) | Latency (us)0.0063 | 27 | |
| Verifier Robustness Evaluation | KernelBench task 001 (held-out) | Hinted ASR62 | 26 | |
| GPU kernel code generation and optimization | KernelBench Level 3 (test) | Correctness Score100 | 18 | |
| GPU kernel code generation and optimization | KernelBench Level 2 (test) | Correctness19 | 18 | |
| GPU kernel code generation and optimization | KernelBench Level 1 (test) | Correctness19 | 18 | |
| GPU kernel optimization | KernelBench Level 3 | Fast132.2 | 17 | |
| GPU kernel optimization | KernelBench Level 1 | Fast1 Score37.2 | 17 | |
| GPU Kernel Optimization | KernelBench Level 2 | Fast1 Score70.6 | 17 | |
| Triton kernel generation | KernelBench LEVEL3 1.0 | Fast1 Score29.8 | 17 | |
| Triton kernel generation | KernelBench LEVEL2 1.0 | Fast1 Score80.9 | 17 | |
| Triton kernel generation | KernelBench LEVEL1 1.0 | Fast139.3 | 17 | |
| Kernel Code Generation | KernelBench Level-3 | Accuracy100 | 14 | |
| Kernel Code Generation | KernelBench Level-2 | Accuracy100 | 14 | |
| Matrix Multiplication | KernelBench FP4 Matmul | Throughput (TF/s)2,898 | 14 | |
| Kernel Generation | KernelBench | Speedup3.45 | 14 | |
| GPU programming | KernelBench level 1 (test) | Compilation Success69 | 12 | |
| GPU kernel generation and optimization | KernelBench Level 2 | Accuracy100 | 11 | |
| GPU kernel generation and optimization | KernelBench Level 1 | Correctness99 | 11 | |
| Kernel Code Generation | KernelBench v0.2.0.dev0 (full 250-task suite) | Correctness Rate98.4 | 11 | |
| Kernel Generation | KernelBench Overall | CR (Round 1)25 | 10 | |
| Kernel Generation | KernelBench Level 2 | CR (Round 1)16 | 10 | |
| Kernel Generation | KernelBench Level 1 | Correctness Rate (Round 1)34 | 10 |