List your company

PDE solver generation on PDE Agent Bench

60HeatAutoPDE
-2.413.83046.2Jun 9, 2026

Evaluation results

MethodLinks
AutoPDEBackbone=GPT 5.12026.066059.522.235.778.953.31005054.5
OpenHandsBackbone=GPT 5.12026.064550014.368.453.371.436.440.3
AutoPDEBackbone=Claude Opus 4.62026.064033.344.446.489.510071.472.754.5
OpenHandsBackbone=Claude Opus 4.62026.0637.505.63.6010042.963.625.7
mini-swe-agentBackbone=GPT 5.12026.0637.538.1035.784.226.785.727.338.2
mini-swe-agentBackbone=Claude Opus 4.62026.06300005.393.3068.222
Naive LLMBackbone=Gemini 2.5 Pro2026.06202.400033.314.34.58.4
Naive LLMBackbone=GPT 5.12026.062023.80068.433.371.4021.5
Naive LLMBackbone=DeepSeek V3.22026.0612.52.45.6002042.906.8
CodePDEBackbone=GPT 5.12026.06514.30036.813.342.94.511
CodePDEBackbone=Claude Opus 4.62026.0609.511.15094.78085.722.731.9
Naive LLMBackbone=Claude Opus 4.62026.060066.742.926.310042.94.525.1