Prompt Optimization on VisEval benchmark leaderboard · SOTA2 Research

Prompt Optimization on VisEval

0.77Accuracy (Easy)PromptAgent

Evaluation Results

Method	Links
PromptAgent2026.02		0.77	0.5	0.39	0.26	0.48
PromptBreeder2026.02		0.76	0.51	0.42	0.35	0.51
CPO2026.02		0.76	0.6	0.49	0.34	0.5475
APE2026.02		0.75	0.59	0.43	0.36	0.5325
Human2026.02		0.74	0.59	0.42	0.26	0.5025
OPRO2026.02		0.74	0.57	0.48	0.34	0.5325
TextGrad2026.02		0.74	0.59	0.42	0.26	0.5125
CoT (1-shot)2026.02		0.73	0.58	0.44	0.27	0.505
Reflexion2026.02		0.73	0.54	0.43	0.29	0.4975
DSPy2026.02		0.58	0.43	0.39	0.13	0.3825