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

Multimodal Reasoning on MathVision

57.9Accuracy

MiMo-VL-7B

Updated 4d ago

Evaluation Results

Method
MiMo-VL-7B 2026.03	57.9	-
MiMo-VL-7B +PC 2026.03	57	-
Qwen2.5-VL-72B w/ RAPID 2025.06	53.4	-
RLSD 2026.04	52.73	-
Qwen2.5-VL-32B w/ RAPID 2025.06	52.1	-
GRPO 2026.04	48.82	-
GRPO + OPSD 2026.04	48.52	-
OPSD 2026.04	47.53	-
Base LLM 2026.04	47.37	-
SDPO 2026.04	47.27	-
Qwen2.5-VL-32B w/ RAPID 2025.06	47	-
GPT-4.1 2025.09	46.4	-
Qwen2.5-VL-7B w/ RAPID 2025.06	43.7	-
Gemini-2.0-Flash 2025.06	43.6	-
OmniCaptioner 2025.06	43.3	-
InternVL3-78B 2025.06	43.1	-
ReVisual-R1-7B 2025.06	43	-
ECSO 2025.06	42.7	-
VL-Rethinker-72B 2025.06	42.2	-
Claude-3.7-Sonnet 2025.06	41.9	-
Claude3.7-Sonnet 2025.09	41.9	-
Qwen2.5-VL-3B w/ RAPID 2025.06	40.8	-
Gemma-3-27B 2025.06	39.8	-
Qwen2.5-VL-72B 2025.06	39.3	-
Qwen2.5-VL-72B 2025.06	39.3	-
Qwen2.5-VL-32B 2025.06	37.8	-
InternVL3-14B 2025.06	37.2	-
MM-Eureka-32B 2025.06	36.6	-
QVQ-72B-Preview 2025.06	34.9	-
InternVL3-38B 2025.06	34.2	-
SRPO-7B 2025.06	32.9	-
DeepSketcher-7B 2025.09	32.3	-
Ovis2-34B 2025.06	31.9	-
Qwen2.5-VL-7B (Bo8) 2025.06	31.6	-
TGRL-DAPO 2026.03	31.26	-
GPT-4o-20241120 2025.06	31.2	-
NoisyRollout 2026.03	30.6	-
MixedR1 7B 2026.01	30.3	-
Gemma-3-12B 2025.06	30.3	-
Ovis2-16B 2025.06	30.1	-
VL-Rethinker-7B 2025.06	30	-
R1-Onevision 7B 2026.01	29.9	-
VisonR1 7B 2026.01	29.9	-
TGRL-GRPO 2026.03	29.87	-
CodeDance 2026.02	29.6	-
InternVL3 2025.11	29.3	-
InternVL3-8B 2025.06	29.3	-
DAPO 2026.03	28.94	-
DeepEyes-v2 2026.02	28.9	-
DeepEyesV2 2025.11	28.9	-
GRPO 2026.02	28.7	-
PyVision-Image 2026.02	28.7	-
TGRL-DAPO 2026.03	28.61	-
Perception-R1 2026.03	28.6	-
Mirage-7B 2025.09	28.6	-
VL-Rethinker 2025.11	28.4	-
GRPO 2026.03	28.32	-
Bagel-Zebra-CoT-7B 2025.09	28.2	-
MM-Eureka 2025.11	28.1	-
LUSPO 2026.02	28	-
MM-Eureka-7B 2025.06	27.9	-
Qwen2.5-VL-7B +PC 2026.03	27.8	-
GSPO 2026.02	27.7	-
Thyme 2026.02	27.6	-
Thyme 2025.11	27.6	-
ThinkLite 2026.03	27.6	-
Mulberry-Qwen-7B 2025.09	27.6	-
Thyme-VL-7B 2025.09	27.6	-
TGRL-GRPO 2026.03	27.41	-
GRPO 2026.03	27.3	-
Qwen2.5-VL-7B 2025.09	27	-
VLAA-Thinker 2026.03	26.9	-
Ours 2026.01	26.7	-
DeepEyes 2026.02	26.6	-
DeepEyes 2025.11	26.6	-
DeepEyes-7B 2025.06	26.6	-
DeepEyes-7B 2025.09	26.6	-
DAPO 2026.03	26.5	-
VLAA-Thinker 7B 2026.01	26.4	-
InternVL3-8B 2025.09	26.3	-
Qwen2.5-VL-7B-Instruct (w/o RLVR) 2026.02	26.2	-
Ours [with DPS] 2026.01	26.1	-
Ours [with DPS and annealing] 2026.01	26.1	-
Ovis2-8B 2025.06	25.9	-
Qwen2.5-VL-7B 2025.06	25.8	-
Qwen2.5-VL-7B 2026.02	25.6	-
Qwen-2.5-VL 2025.11	25.6	-
Qwen2.5-VL-7B 2026.03	25.1	-
OpenVLThinker 2026.03	25	-
Qwen-2.5-VL-7B 2026.03	25	-
VILASR-7B 2025.09	25	-
ThinkLite 2025.11	24.6	-
VLAA-Thinker 2025.11	24.2	-
Qwen2.5-VL-3B +PC 2026.03	23.3	-
GPT-4V 2026.01	22.76	-
R1-Onevision 2026.03	22.5	-
MathCoder-VL 2025.09	22.3	-
InternVL2.5 8B 2026.01	22	-
Qwen2.5-VL-3B 2025.06	21.9	-
Qwen2.5-VL-3B 2026.03	21.2	-

Showing 100 of 125 rows