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

Multimodal Reasoning on MMMU Pro (Accuracy)

76.96Accuracy

Gemini-2.5 (Pro)

Updated 3d ago

Evaluation Results

Method	Links
Gemini-2.5 (Pro) 2026.01		76.96
Qwen3-VL (Thinking) 2026.01		72.37
Seed-1.5-VL (Thinking) 2026.01		70.6
STEP3-VL-10B (PaCoRe) 2026.01		67.18
GLM-4.6V 2026.01		65.84
STEP3-VL-10B (SeRe) 2026.01		64.08
GPT-4o 2026.01		56.13
GPT-4o 2026.01		51.9
Gemini-1.5-Pro 2026.01		51.47
Qwen3-VL-8B-Instruct + CARE 2025.12		46.7
MiMo-VL-7B-RL 2025.12		46.2
MiMo-VL-7B-SFT 2025.12		45.2
Qwen3-VL-8B-Instruct + CARE 2025.12		41.7
Ours [with DPS and annealing] 2026.01		41
Ours [with DPS] 2026.01		40.7
Ours 2026.01		40.6
MiMo-VL-7B-RL 2025.12		40.3
Qwen2.5-VL-7B + CARE 2025.12		39.7
VLAA-Thinker 7B 2026.01		39.5
MiMo-VL-7B-SFT 2025.12		39.4
Qwen2.5-VL-7B + GSPO 2025.12		38.9
Qwen2.5VL 7B 2026.01		38
MixedR1 7B 2026.01		38
LLaVA-OneVision-1.5 8B 2025.12		37.4
Qwen2.5-VL-7B + DAPO 2025.12		37.3
Qwen2.5-VL-7B + CARE 2025.12		37.1
Qwen2.5-VL-7B + GRPO 2025.12		36.4
Qwen2.5-VL-7B + GSPO 2025.12		36.4
Qwen2.5-VL-7B 2025.12		36.3
Qwen2.5-VL-7B + DAPO 2025.12		34.7
InternVL2.5 8B 2026.01		34.3
R1-Onevision 7B 2026.01		33.9
Qwen2.5-VL-3B + CARE 2025.12		33.1
Qwen2.5-VL-7B 2025.12		32.8
Qwen2.5-VL-7B + GRPO 2025.12		32.8
Qwen2.5-VL-3B 2025.12		31.1
Qwen2-VL 2026.01		30.5
VisonR1 7B 2026.01		30.3
InternVL2 2026.01		29
Qwen3-VL 2025.12		28.1
NVILA 2026.01		27.8
LinMU-NV 2026.01		27.3
LLaVA-OneVision-1.5 8B 2025.12		25.2
LLaVA-OV 2026.01		24.1
Qwen2.5-VL-3B + CARE 2025.12		24
HyperVL 2025.12		23.9
HyperVL ViTL 2025.12		23.8
InternVL3.5 2025.12		23.6
SAIL-VL2 2025.12		22.2
AndesVL 2025.12		22
Qwen2.5-VL-3B 2025.12		21.3
Qwen2-VL 2025.12		19.9
Ovis2 2025.12		18.5
InternVL3 2025.12		17.5
SAIL-VL1.5 2025.12		16.4