List your company

Code Generation on LiveCodeBench v5 (held-out)

46.8Pass@1HSD
31.40835.40439.443.396Jun 14, 2026

Evaluation results

MethodLinks
HSDBase Model=Qwen3-32B-Base2026.0646.8
GRPO + OPSDBase Model=Qwen3-32B-Base2026.0644.1
RLSDBase Model=Qwen3-32B-Base2026.0643.9
SDPOBase Model=Qwen3-32B-Base2026.0643.7
OPSDBase Model=Qwen3-32B-Base2026.0643
DAPOBase Model=Qwen3-32B-Base2026.0642.9
GSPOBase Model=Qwen3-32B-Base2026.0642.6
Dr. GRPOBase Model=Qwen3-32B-Base2026.0642.2
GRPOBase Model=Qwen3-32B-Base2026.0641.6
HSDBase Model=Qwen3-8B-Base2026.0637.2
GRPO + OPSDBase Model=Qwen3-8B-Base2026.0635.1
RLSDBase Model=Qwen3-8B-Base2026.0634.8
SDPOBase Model=Qwen3-8B-Base2026.0634.5
OPSDBase Model=Qwen3-8B-Base2026.0633.7
DAPOBase Model=Qwen3-8B-Base2026.0633.4
GSPOBase Model=Qwen3-8B-Base2026.0633
Dr. GRPOBase Model=Qwen3-8B-Base2026.0632.6
GRPOBase Model=Qwen3-8B-Base2026.0632