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

Multimodal Reasoning on DynaMath

67.2Accuracy

SwimBird

Updated 7d ago

Evaluation Results

Method
SwimBird 2026.02	67.2	-	-
Qwen3-VL-8B-Instruct 2026.02	65.3	-	-
PyVision-Image 2026.02	61.6	-	-
DeepEyesV2 2026.02	57.2	-	-
DeepEyes-v2 2026.02	57.2	-	-
DeepEyesV2 2025.11	57.2	-	-
ReLaX-VL-7B 2025.12	55.9	-	-
VL-Rethinker-7B 2025.12	55.2	-	-
DeepEyes 2026.02	55	-	-
DeepEyes 2026.02	55	-	-
DeepEyes 2025.11	55	-	-
MM-Eureka-7B 2025.12	54.4	-	-
Qwen2.5-VL-7B-Instruct 2026.02	53.3	-	-
Qwen2.5-VL-7B 2026.02	53.3	-	-
Qwen-2.5-VL 2025.11	53.3	-	-
Qwen2.5-VL-7B 2025.12	53.2	-	-
ReLaX-VL-3B 2025.12	52.2	-	-
Vision-R1-7B 2025.12	52	-	-
R1-VL-7B 2025.12	45.2	-	-
Athena-PRM 2025.06	42.5	-	-
Qwen2-VL-7B 2025.12	42.1	-	-
Gemini-2.0-Flash 2025.06	42.1	-	-
Qwen2.5-VL-3B 2025.12	40	-	-
Intern2-VL-8B 2025.12	39.7	-	-
Claude-3.7-Sonnet 2025.06	39.7	-	-
Athena-ORM 2025.06	39.6	-	-
OpenVLThinker-7B 2025.12	38.6	-	-
Qwen2.5-VL-32B w/ RAPID 2025.06	38.3	-	-
Qwen2.5-VL-72B w/ RAPID 2025.06	37.9	-	-
Self-consistency 2025.06	37.6	-	-
Qwen2.5-VL-32B w/ RAPID 2025.06	36.5	-	-
Qwen2.5-VL-72B 2025.06	35.9	-	-
Qwen2.5-VL-72B 2025.06	35.9	-	-
Qwen2.5-VL-72B 2025.06	35.9	-	-
InternVL3-38B 2025.06	35.3	-	-
InternVL3-78B 2025.06	35.1	-	-
VL-Rethinker-72B 2025.06	34.9	-	-
GPT-4o-20241120 2025.06	34.5	-	-
MM-Eureka-32B 2025.06	33.5	-	-
Qwen2.5-VL-32B 2025.06	33.3	-	-
Qwen2.5-VL-7B w/ RAPID 2025.06	32.7	-	-
InternVL3-14B 2025.06	31.3	-	-
QVQ-72B-Preview 2025.06	30.7	-	-
ReVisual-R1-7B 2025.06	30.5	-	-
OmniCaptioner 2025.06	30.5	-	-
Qwen2.5-VL-7B (Bo8) 2025.06	29.7	-	-
Qwen2.5-VL-3B w/ RAPID 2025.06	29.3	-	-
Gemma-3-27B 2025.06	28.5	-	-
Ovis2-34B 2025.06	27.5	-	-
Ovis2-16B 2025.06	26.3	-	-
InternVL3-8B 2025.06	25.5	-	-
ECSO 2025.06	25	-	-
Athena-PRM 2025.06	23.4	-	-
Athena-ORM 2025.06	23.1	-	-
VisualPRM-8B 2025.06	23	-	-
Self-consistency 2025.06	22.9	-	-
MM-Eureka-7B 2025.06	22.6	-	-
Qwen2.5-VL-7B 2025.06	21.8	-	-
VL-Rethinker-7B 2025.06	21.4	-	-
Gemma-3-12B 2025.06	20.8	-	-
Ovis2-8B 2025.06	20.4	-	-
Qwen2.5-VL-7B 2025.06	19.4	-	-
Athena-PRM 2025.06	18.7	-	-
VisualPRM-8B 2025.06	18	-	-
Athena-ORM 2025.06	15.2	-	-
Self-consistency 2025.06	13.8	-	-
Qwen2.5-VL-3B 2025.06	13.4	-	-
InternVL2.5-8B 2025.06	9.4	-	-
GRPO 2026.02	0.262	-	-
GSPO 2026.02	0.259	-	-
LUSPO 2026.02	0.246	-	-
Qwen2.5-VL-7B-Instruct (w/o RLVR) 2026.02	0.202	-	-
GPT-4o-mini 2025.08	-	40.35	37.46
Gemini-2.0-Flash 2025.08	-	55.79	43.25
InternVL2.5-8B 2025.08	-	48.88	51.24
Qwen2.5-VL-7B 2025.08	-	52.81	52.89
Qwen2.5-VL-72B 2025.08	-	51.75	53.25
Qwen3-VL-4B-Thinking 2025.08	-	55.79	62.82
MiMo-VL-7B 2025.08	-	60.5	66.81
VisualPRM-8B 2025.08	-	30	62.08
VRPRM 2025.08	-	53.51	67.95
VRPRM w/o CoT 2025.08	-	41.05	53.06
VRPRM w/o RL 2025.08	-	52.46	63.08
VRPRM w/o RL & w/o CoT 2025.08	-	50.18	55.26
VRPRM-MiMo 2025.08	-	59.47	74.55
VRPRM-MiMo w/o CoT 2025.08	-	59.12	73.2
VRPRM-MiMo w/o RL 2025.08	-	60.04	61.67
VRPRM-MiMo w/o RL & w/o CoT 2025.08	-	59.65	50.48
VRPRM-Qwen3 2025.08	-	57.02	72.76
VRPRM-Qwen3 w/o CoT 2025.08	-	53.33	71.85
VRPRM-Qwen3 w/o RL 2025.08	-	59.82	57.58
VRPRM-Qwen3 w/o RL & w/o CoT 2025.08	-	55.61	56.23