List your company

Hallucination Evaluation on MMHal

4.2ScoreLLaVA-OneVision-7B w/ UnifiedReward-Think
1.76642.39823.033.6618Dec 6, 2024Mar 11, 2025Jun 14, 2025Sep 17, 2025Dec 21, 2025Mar 26, 2026Jun 30, 2026

Evaluation results

MethodLinks
LLaVA-OneVision-7B w/ UnifiedReward-ThinkReward Model Paradigm=Generative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.044.2—————
LLaVA-OneVision-7B w/ Skywork-VLReward Model Paradigm=Discriminative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.044.2—————
LLaVA-OneVision-7B w/ VL-MDRReward Model Paradigm=Discriminative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.044.2—————
LLaVA-OneVision-7B w/ IXC-2.5-RewardReward Model Paradigm=Discriminative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.044.1—————
LLaVA-OneVision-7B w/ MM-RLHF-RewardReward Model Paradigm=Discriminative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.044.1—————
CycleCapT2I Model=SD32026.034.02—————
CycleCapT2I Model=FLUX.1-dev2026.034.02—————
GPT-4o-20240513Model Name=GPT-4o-202405132024.124—————
LLaVA-OneVision-7B w/ UnifiedRewardReward Model Paradigm=Generative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.044—————
CyclePrefT2I Model=SD32026.033.95—————
RICO-FlashT2I Model=FLUX.1-dev2026.033.92—————
LLaVA-OneVision-7B w/ LLaVA-CriticReward Model Paradigm=Generative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.043.9—————
InternVL2.5-78BModel Name=InternVL2.5-78B2024.123.89—————
Baseline (Qwen2-VL-7B)T2I Model=—2026.033.85—————
InternVL2-Llama3-76BModel Name=InternVL2-Llama3-76B2024.123.83—————
ADAPT-TFBackbone=Qwen2.5-VL-7B2026.063.83————31
InternVL2-40BModel Name=InternVL2-40B2024.123.75—————
ADAPTBackbone=Qwen2.5-VL-7B2026.063.75————30
InternVL2.5-38BModel Name=InternVL2.5-38B2024.123.71—————
InternVL2.5-26BModel Name=InternVL2.5-26B2024.123.7—————
InternVL2.5-8BModel Name=InternVL2.5-8B2024.123.65—————
MiniCPM-V2.6Model Name=MiniCPM-V2.62024.123.6—————
DPOBackbone=Qwen2.5-VL-7B2026.063.59————39
InternVL2-26BModel Name=InternVL2-26B2024.123.55—————
Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.063.55————40
VCDBackbone=Qwen2.5-VL-7B2026.063.42————40
Qwen2-VL-7BModel Name=Qwen2-VL-7B2024.123.4—————
InternVL2-8BModel Name=InternVL2-8B2024.123.33—————
InternVL2.5-4BModel Name=InternVL2.5-4B2024.123.31—————
LLaVA-OneVision-7BBase Model=LLaVA-OneVision-7B2026.043.2—————
ADAPTBackbone=Qwen2.5-VL-3B2026.063.16————37
VCDBackbone=Qwen2.5-VL-3B2026.063.14————42
InternVL-Chat-V1.5Model Name=InternVL-Chat-V1.52024.123.11—————
ADAPT-TFBackbone=Qwen2.5-VL-3B2026.063.11————39
InternVL2.5-2BModel Name=InternVL2.5-2B2024.122.94—————
TARS (Replace)Backbone=Muffin-13B2025.072.76——43——
InternVL2-4BModel Name=InternVL2-4B2024.122.75—————
TARS (Mask)Backbone=Muffin-13B2025.072.75——41——
CHiP-DPOBackbone=Muffin-13B2025.072.7——47——
ADAPTBackbone=LLaVA-v1.5-13B2026.062.64————49
ADAPTBackbone=LLaVA-v1.5-7B2026.062.62————53
Qwen2.5-VL-3BBackbone=Qwen2.5-VL-3B2026.062.62————49
ADAPT-TFBackbone=LLaVA-v1.5-7B2026.062.55————53
ADAPT-TFBackbone=LLaVA-v1.5-13B2026.062.55————51
DPOBackbone=Qwen2.5-VL-3B2026.062.53————52
InternVL2-2BModel Name=InternVL2-2B2024.122.52—————
RLHF-VBackbone=LLaVA-v1.5-13B2026.062.5————52
InternVL2.5-1BModel Name=InternVL2.5-1B2024.122.49—————
DPOBackbone=LLaVA-v1.5-13B2026.062.47————51
OPERABackbone=LLaVA-v1.5-13B2026.062.46————52
DPOBackbone=Muffin-13B2025.072.45——55——
mDPOBackbone=LLaVA-v1.5-7B2026.062.39————54
LLaVA-v1.5-13BBackbone=LLaVA-v1.5-13B2026.062.38————53
VCDBackbone=LLaVA-v1.5-13B2026.062.31————54
SIMABackbone=LLaVA-v1.5-7B2026.062.3—————
Muffin-13BBackbone=Muffin-13B2025.072.27——58——
InternVL2-1BModel Name=InternVL2-1B2024.122.25—————
OPERABackbone=LLaVA-v1.5-7B2026.062.15————54
DPOBackbone=LLaVA-v1.5-7B2026.062.14————65
RLHFBackbone=Muffin-13B2025.072.12——64——
VCDBackbone=LLaVA-v1.5-7B2026.062.12————54
LLaVA-v1.5-7BBackbone=LLaVA-v1.5-7B2026.061.86————64
Gemini 2.5 FlashSize=-, mode=instruct2026.04————4.623.9
InternVL3.5Size=8B, mode=instruct2026.04————3.834.7
LLaVA2025.04—31.970.8———
MiniCPM-o 4.5Size=9B, mode=instruct2026.04————4.724.3
POPEN2025.04—42.147.9———
Qwen3-OmniSize=30B-A3B, mode=instruct2026.04————4.631.6
Qwen3-VLSize=8B, mode=instruct2026.04————4.729.9
RLHF-V2025.04—4052.1———
SFT + DA-DPOBase Model=LLaVA-OneVision-7B [3]2026.06—————2.78
SFT + DA-DPOBase Model=InternVL2.5-26B [4]2026.06—————2.4
SFT + DPOBase Model=LLaVA-OneVision-7B [3]2026.06—————2.61
SFT + DPOBase Model=InternVL2.5-26B [4]2026.06—————2.35
SFT + VIGIL (Ours)Base Model=LLaVA-OneVision-7B [3]2026.06—————2.45
SFT + VIGIL (Ours)Base Model=InternVL2.5-26B [4]2026.06—————2.15
SFT onlyBase Model=LLaVA-OneVision-7B [3]2026.06—————2.76
SFT onlyBase Model=InternVL2.5-26B [4]2026.06—————2.5