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

Subjective Search Question Answering on Internal Evaluation Subjective Q&A

190V4 Win Count

DeepSeek-V4-Pro

Updated 1mo ago

Evaluation Results

Method	Links
DeepSeek-V4-Pro 2026.04		190	74	403	28.5	11.1	60.4
DeepSeek-V3.2 2026.04		190	74	403	28.5	11.1	60.4
DeepSeek-V4-Pro 2026.04		32	11	49	34.8	12	53.3
DeepSeek-V3.2 2026.04		32	11	49	34.8	12	53.3
DeepSeek-V4-Pro 2026.04		30	8	58	31.2	8.3	60.4
DeepSeek-V3.2 2026.04		30	8	58	31.2	8.3	60.4
DeepSeek-V4-Pro 2026.04		28	5	67	28	5	67
DeepSeek-V3.2 2026.04		28	5	67	28	5	67
DeepSeek-V4-Pro 2026.04		28	20	48	29.2	20.8	50
DeepSeek-V3.2 2026.04		28	20	48	29.2	20.8	50
DeepSeek-V4-Pro 2026.04		26	19	50	27.4	20	52.6
DeepSeek-V3.2 2026.04		26	19	50	27.4	20	52.6
DeepSeek-V4-Pro 2026.04		23	8	61	25	8.7	66.3
DeepSeek-V3.2 2026.04		23	8	61	25	8.7	66.3
DeepSeek-V4-Pro 2026.04		23	3	70	24	3.1	72.9
DeepSeek-V3.2 2026.04		23	3	70	24	3.1	72.9