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

Multi-image Understanding on MMIU

55.8Accuracy

InternVL2.5-78B

Updated 4d ago

Evaluation Results

Method	Links
InternVL2.5-78B 2024.12		55.8
GPT-4o 2026.01		55.7
GPT-4o-20240513 2024.12		55.7
InternVL2.5-38B 2024.12		55.3
GeM-VG 2026.01		55.01
Migician 2026.01		54.89
Qwen2-VL-7B 2026.01		54.36
Claude-3.5-Sonnet 2024.12		53.4
Gemini-1.5-Pro 2024.12		53.4
DAPO (Ours) 2026.01		53.3
UniVG-R1 2026.01		53.03
VISC 2026.01		52.8
Qwen2VL-7B 2026.01		52.6
DPS (Ours) 2026.01		52.2
Two-stage RL (Ours) 2026.01		52.2
Qwen2.5-VL 2026.01		50.6
MiniCPM-V 2.6 2026.01		50.19
Ours 2025.07		49.93
InternVL2.5-26B 2024.12		49.4
Qwen2.5-VL-7B + SFT 2025.07		48.21
InternVL2-40B 2024.12		47.9
Qwen2.5-VL-7B 2025.07		47.52
LLaVA-NeXT-Interleave 2026.01		47.3
InternVL2.5 2026.01		46.7
InternVL2.5-8B 2024.12		46.7
Mantis-Idefics2 2026.01		45.6
Mantis 2026.01		45.6
Ours (masked) (LLaVA-OV-7B) 2026.01		45.5
LLaVA-OV-7B 2026.01		45
TW-GRPO 2026.01		44.9
VideoRFT 2026.01		44.5
LLaVA-OV-7B 2026.01		44.46
InternVL2-Llama3-76B 2026.01		44.2
InternVL2-Llama3-76B 2024.12		44.2
InternVL2.5-4B 2024.12		43.8
InternVL2.5-2B 2024.12		43.5
InternVL2-4B 2024.12		43.3
InternVL2-26B 2024.12		42.6
InternVL2-8B 2026.01		42
InternVL2-8B 2024.12		42
LLaVA-OneVision 2026.01		40.3
InternVL2-2B 2024.12		39.8
mPLUG-Owl3 2026.01		39.7
Ours (LLaVA-OV-1.5B) 2026.01		38.9
Qwen2VL-2B 2026.01		38.7
InternVL2.5-1B 2024.12		38.5
InternVL-Chat-V1.5 2024.12		37.4
InternVL2-1B 2024.12		37.3
Ours (LLaVA-OV-0.5B) 2026.01		37.2
InternVL2-8B 2026.01		36.8
Ours (masked) (LLaVA-OV-0.5B) 2026.01		36.3
Ours (masked) (LLaVA-OV-1.5B) 2026.01		35.5
VISOR 2026.03		34.5
LLaVA-OV-0.5B 2026.01		34.2
LLaVA 2026.03		34.2
LLaVA-OV-1.5B 2026.01		33.4
VILA1.5 2026.01		32.5
Idefics2-8B 2026.01		27.8
OpenFlamingo-v2 2026.01		23.7
CogVLM 2026.01		23.57
LLaVA 1.6 2026.01		22.2
mPLUG-Owl3 2026.01		21.72
LLaVA-v1.5-7B 2026.01		19.2
LLaVA-1.5 2026.01		19.2
InternVL2-2B 2026.01		13.6