Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Streaming Evaluation on SVBench

57.99SA Score

GPT-4o

33.34239.74146.1452.539Jun 16, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
57.9963.5267.7260.1859.2559.9761.27
2026.06
56.3761.4165.859.1857.1657.9359.12
2026.06
52.195962.8558.3554.9555.8757.41
2026.06
48.2156.1260.3156.3352.4652.8455.21
2026.06
46.4452.7358.3553.4848.3249.6753.15
2026.06
44.3355.4959.8352.8750.4850.6852.2
2026.06
43.5651.5255.7150.7947.7748.1549.76
2026.06
41.7649.7253.2548.4447.146.5849.16
2026.06
40.4448.3452.8446.9345.145.0446.89
2026.06
34.2939.6847.6535.3336.6836.1238.76