List your company

Code generation on MooseBench standard non-registry, 220-case

81.6IFSPDE-Refine
59.03264.89170.7576.609May 10, 2026

Evaluation results

MethodLinks
PDE-RefineLLM=Claude Sonnet 4.62026.0581.6
PDE-RefineLLM=GPT-5.42026.0580.2
SpecGenLLM=Claude Sonnet 4.62026.0580
PDE-RefineLLM=DeepSeek V4 Flash2026.0578.2
SpecGenLLM=GPT-5.42026.0576.9
DirectLLM=GPT-5.42026.0575.6
DirectLLM=Claude Sonnet 4.62026.0574.4
SpecGenLLM=DeepSeek V4 Flash2026.0566.8
DirectLLM=DeepSeek V4 Flash2026.0559.9