List your company

Code Generation on HumanEval (held-out)

95.1AccuracySC-Route
57.6667.3877.186.82Jun 22, 2026

Evaluation results

MethodLinks
SC-RouteBackbone=Qwen3-32B2026.0695.1——
DARTBackbone=Qwen3-32B2026.0695.16376.8
NTBackbone=Qwen3-32B2026.0679.9——
SC-RouteBackbone=Qwen3-8B2026.0678.7——
DARTBackbone=Qwen3-8B2026.0678.75554.9
SC-RouteBackbone=Qwen3-14B2026.0678.7——
DARTBackbone=Qwen3-14B2026.0678.76068.9
ATBackbone=Qwen3-32B2026.0672.6——
NTBackbone=Qwen3-14B2026.0671.3——
ATBackbone=Qwen3-14B2026.0666.5——
NTBackbone=Qwen3-8B2026.0660.4——
ATBackbone=Qwen3-8B2026.0659.1——