List your company

Speaker-attribute classification on WhoSaidIt re-annotated (test)

91.9Male AccuracyGPT 4.1
74.01278.65683.387.944May 25, 2026

Evaluation results

MethodLinks
GPT 4.12026.0591.994.16986.591.592.991.785.775.3
Claude 3.7 Sonnet2026.0591.996.57086.894.592.490.390.978.6
Gemini 2.5 Flash2026.0590.988.366.581.681.283.990.192.869.5
DeepSeek V32026.0586.282.366.984.184.588.189.978.274.1
GPT 4.1prompting=with rationales2026.058588.942.58383.586.685.772.963.2
Claude 3.7 Sonnetprompting=with rationales2026.0584.893.343.983.389.485.583.48366.1
Gemini 2.5 Flashprompting=with rationales2026.0583.277.940.477.56470.18386.657.9
DeepSeek V3prompting=with rationales2026.0574.76738.579.470.277.782.558.661.7