Multimodal-to-text retrieval on MM-BRIGHT

45.4Acad ScoreHIVE

Evaluation Results

Method
HIVE2026.04	45.4	—	42.7	42.5	25.1	45.2	44.6	40.1	34.8	42	36.1	53.6	37.3	22.5	14.9	39.1	44.9	49.2	41.9	30.1	64.6	49.2	41	40.1	68.2	36.5	41.9	49.4	51.8	34.2	41.7
MARVEL2026.04	36.8	—	42.2	40.3	24.8	39	44.5	32.9	28.4	41.1	21.3	56.4	36.9	20.4	11.3	33.8	50.7	40	37.7	31.9	53.6	47.1	41.1	31.4	65.5	26.9	49.3	41.8	54.1	18.8	37.9
BRIDGE2026.04	27.7	—	41.1	34.3	20.7	31.9	32.3	26.4	30.8	31.3	21.3	50.4	28	31.4	8.9	19	30.1	23.1	28.5	18	40.2	28.9	22.6	29	45.2	34.6	27.1	28.3	45.7	23.6	29.7
GME-7B2026.04	27.6	—	15.2	21.9	14	9.3	26.2	13.4	19.2	19	17	34.2	19.6	7.1	5.6	18.7	47.3	12.6	18.6	18	35	26.5	32	17	43.9	15.6	33.2	25.6	30.8	15.3	22
GME-7B2026.04	27.6	—	15.2	21.9	14	9.3	26.2	13.4	19.2	19	17	34.2	19.6	7.1	5.6	18.7	47.3	12.6	18.6	18	35	26.5	32	17	43.9	15.6	33.2	25.6	30.8	15.3	22
GME-7B2026.04	27.6	—	15.2	21.9	14	9.3	26.2	13.4	19.2	19	17	34.2	19.6	7.1	5.6	18.7	47.3	12.6	18.6	18	35	26.5	32	17	43.9	15.6	33.2	25.6	30.8	15.3	22
NOMIC2026.04	22.6	—	26.9	30.6	17.2	34	30.1	23.4	33.8	33.9	28.7	34.3	22.7	22.4	12.1	30.3	26.2	21.1	23.9	21.7	47.6	30.9	28.9	24.1	43.1	25.8	27.6	24.7	36.7	16.2	27.6
NOMIC2026.04	22.6	—	26.9	30.6	17.2	34	30.1	23.4	33.8	33.9	28.7	34.3	22.7	22.4	12.1	30.3	26.2	21.1	23.9	21.7	47.6	30.9	28.9	24.1	43.1	25.8	27.6	24.7	36.7	16.2	27.6
NOMIC2026.04	22.6	—	26.9	30.6	17.2	34	30.1	23.4	33.8	33.9	28.7	34.3	22.7	22.4	12.1	30.3	26.2	21.1	23.9	21.7	47.6	30.9	28.9	24.1	43.1	25.8	27.6	24.7	36.7	16.2	27.6
JINA-CLIP2026.04	22.3	—	20.5	30.6	14.4	27	24.6	19.4	23.7	26.8	24.3	26.1	22.6	15.5	10.8	19	32.3	13.5	20.8	19.4	35.3	21	24.3	24.3	45.6	20.3	20.5	24.3	26.6	11.6	23
JINA-CLIP2026.04	22.3	—	20.5	30.6	14.4	27	24.6	19.4	23.7	26.8	24.3	26.1	22.6	15.5	10.8	19	32.3	13.5	20.8	19.4	35.3	21	24.3	24.3	45.6	20.3	20.5	24.3	26.6	11.6	23
JINA-CLIP2026.04	22.3	—	20.5	30.6	14.4	27	24.6	19.4	23.7	26.8	24.3	26.1	22.6	15.5	10.8	19	32.3	13.5	20.8	19.4	35.3	21	24.3	24.3	45.6	20.3	20.5	24.3	26.6	11.6	23
GME-2B2026.04	16.2	—	22.9	27.2	13.3	16.4	20.5	10.5	21.1	22.7	23.9	25.9	18.2	9.8	5.9	15.8	31.1	10	15.6	15.2	30.7	20	25.8	16.2	41.6	15.5	21.9	16.7	23.9	12.4	19.5
GME-2B2026.04	16.2	—	22.9	27.2	13.3	16.4	20.5	10.5	21.1	22.7	23.9	25.9	18.2	9.8	5.9	15.8	31.1	10	15.6	15.2	30.7	20	25.8	16.2	41.6	15.5	21.9	16.7	23.9	12.4	19.5
GME-2B2026.04	16.2	—	22.9	27.2	13.3	16.4	20.5	10.5	21.1	22.7	23.9	25.9	18.2	9.8	5.9	15.8	31.1	10	15.6	15.2	30.7	20	25.8	16.2	41.6	15.5	21.9	16.7	23.9	12.4	19.5
CLIP2026.04	4.8	—	14.8	9.6	6.1	17.9	10.9	11.4	9.4	9.8	12.3	5.5	8.3	14.8	2.6	10.6	2.3	6	8.7	5.4	19.7	15	10.7	15.4	19.1	13.1	8.9	9	16.1	2.1	10.4
CLIP2026.04	4.8	—	14.8	9.6	6.1	17.9	10.9	11.4	9.4	9.8	12.3	5.5	8.3	14.8	2.6	10.6	2.3	6	8.7	5.4	19.7	15	10.7	15.4	19.1	13.1	8.9	9	16.1	2.1	10.4
CLIP2026.04	4.8	—	14.8	9.6	6.1	17.9	10.9	11.4	9.4	9.8	12.3	5.5	8.3	14.8	2.6	10.6	2.3	6	8.7	5.4	19.7	15	10.7	15.4	19.1	13.1	8.9	9	16.1	2.1	10.4
BGE-VL2026.04	4.2	—	5.7	10.8	6.8	13.1	10.1	13.3	11.6	12.6	7.2	11.6	8.9	11.3	4.5	16.1	14.2	9.5	6.4	2.4	10.2	8.9	12	9.6	17.5	13.8	8.6	10.1	10.1	8.1	10
BGE-VL2026.04	4.2	—	5.7	10.8	6.8	13.1	10.1	13.3	11.6	12.6	7.2	11.6	8.9	11.3	4.5	16.1	14.2	9.5	6.4	2.4	10.2	8.9	12	9.6	17.5	13.8	8.6	10.1	10.1	8.1	10
BGE-VL2026.04	4.2	—	5.7	10.8	6.8	13.1	10.1	13.3	11.6	12.6	7.2	11.6	8.9	11.3	4.5	16.1	14.2	9.5	6.4	2.4	10.2	8.9	12	9.6	17.5	13.8	8.6	10.1	10.1	8.1	10
SIGLIP2026.04	3.6	—	11.9	11.6	7.3	15.3	11.8	14.8	16.8	9.1	4.4	12.6	10	10.2	2.6	14.3	6.5	9.8	7.9	7	16.4	13	6.5	9.2	21.4	16.5	12.4	11.5	13.1	5.8	10.8
SIGLIP2026.04	3.6	—	11.9	11.6	7.3	15.3	11.8	14.8	16.8	9.1	4.4	12.6	10	10.2	2.6	14.3	6.5	9.8	7.9	7	16.4	13	6.5	9.2	21.4	16.5	12.4	11.5	13.1	5.8	10.8
SIGLIP2026.04	3.6	—	11.9	11.6	7.3	15.3	11.8	14.8	16.8	9.1	4.4	12.6	10	10.2	2.6	14.3	6.5	9.8	7.9	7	16.4	13	6.5	9.2	21.4	16.5	12.4	11.5	13.1	5.8	10.8