AgenticPICBenchmark
Menu

PICBench

PICBench: Photonic Integrated Circuit benchmark

Published tasks07
Model–agent pair01
Latest runAUG 11, 2026
Benchmark comparison

Final performance by task

Normalized official score · 0–1 scale · higher is better

CodexGPT-5.5
Final normalized performance by benchmark task
TaskAgentModelFinal score
WDM-v1codexgpt-5.50.0507
Microresonator-v1codexgpt-5.50.8909
Microresonator-v2codexgpt-5.50.0000
WDM-v2codexgpt-5.50.0000
MZI-v1codexgpt-5.50.0000
Quantum-4codexgpt-5.50.1073
Quantum-8codexgpt-5.50.0000