Our new X account is live! Follow @wizwand_team for updates

Referring Expression Comprehension on RefBench PRO

71.44Acc (Phrase)

Qwen3-VL-8B

Updated 4d ago

Evaluation Results

Method
Qwen3-VL-8B 2025.12	71.44	62.16	-	-	-	-	-	-	-	-
Qwen3-VL 2025.12	71.4	62.2	76.6	76.1	67.3	73.3	68.9	68.3	15.8	68.6
Qwen2.5-VL 2025.12	66.7	59.5	68.6	69.1	69.4	69.1	61.6	64.8	23.6	63.2
Rex-Thinker 2025.12	63.6	53	67.1	64.5	61.7	64.4	59.3	65.6	-	62.4
Ovis2.5-9B 2025.12	61.74	51.45	-	-	-	-	-	-	-	-
Ovis2.5 2025.12	61.7	51.5	65.7	63.6	59.7	63	58.7	61	-	59.9
GLM-4.1V-Base 2025.12	60.1	50.1	62.9	61	57.7	60.5	57	61.9	-	59.4
Qwen2.5-VL 2025.12	57.6	48.5	61.7	63	58.6	61.1	49.1	55.6	3.1	52.3
Qwen2.5-VL-7B 2025.12	57.58	48.5	-	-	-	-	-	-	-	-
CogVLM-Grounding 2025.12	57.1	47.5	62.4	62.4	55.9	60.2	49.4	55.2	-	52.3
CogVLM-Grounding-17B 2025.12	57.05	47.54	-	-	-	-	-	-	-	-
LLaVA-OneVision 2025.12	56.5	47.1	60.1	59.4	53.7	57.7	54.2	55	-	54.6
Mimo-VL-RL 2025.12	56.3	46.9	60.9	58.4	57.3	58.9	51.8	52.9	0.1	52.4
VLM-R1-3B 2025.12	54.41	45.34	-	-	-	-	-	-	-	-
VLM-R1 2025.12	54.4	45.3	59	58	54.1	57	47.8	53.2	-	50.5
UniVG-R1 2025.12	53	44.2	59.4	57.2	55.1	57.2	48.3	44.9	-	46.6
Migician 2025.12	52.3	43.6	57.3	59.7	52.8	56.6	45.4	46.1	-	45.7
LLaVA-OneVision-1.5 2025.12	50.7	42.3	54.5	54	48.4	52.3	48.1	48.6	-	48.3
ChatRex 2025.12	49.5	41.3	54.7	51.1	53.2	53	45.1	43.4	-	44.2
Qwen2-VL 2025.12	45.4	42.6	55.3	47.8	37.8	47	39.4	46.5	28.5	43
InternVL3.5 2025.12	41.5	34.6	45.7	41.2	45.3	44.1	37.8	37.3	-	37.5
Grounding DINO L 2025.12	37.6	31.3	47.5	43.3	31.8	40.9	35	30.3	0.1	32.7
Grounding Dino L 2025.12	37.58	31.34	-	-	-	-	-	-	-	-
GLEE 2025.12	36.1	31.2	48.2	38.4	34.5	40.4	31.4	27.9	7.1	29.7
PaDT 2025.12	26.6	22.2	30.4	28	30.4	29.6	23.7	20.8	-	22.3
GPT-5 2025.12	26.1	21.8	29.2	25.5	27	27.2	26.2	22.9	-	24.6
InternVL3 2025.12	21.8	22.3	35	24.9	28.2	29.4	24.3	20.8	24.8	22.5
InternVL3 2025.12	20.1	20.3	24.9	18.4	22.3	21.9	19.8	15	21.3	17.4
GPT-4o 2025.12	12.1	10.1	11.7	12.8	11.9	12.1	12.4	11.6	-	12
Gemini-2.5-pro 2025.12	9.6	8	10.4	11.5	10.8	10.9	7.2	8.2	-	7.7