AI Paper Atlas

Research map / Benchmarks / MMHal-Bench

MMHal-Bench leaderboard

23 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on MMHal-Bench. Best Score: 4.26 (VAR, ICLR 2025).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

MMHal-Bench · Score

23 papers, each with its best reported number for its own method; higher is better.

#Score ↑MethodPaperTopic
14.26VAR
LLaVA-1.5-7B
See What You Are Told: Visual Attention Sink in Large Multimodal Models
ICLR 2025 · Visual hallucination results t
Multimodal LLMs
23.7Qwen2VL-7B + IPA
Qwen2VL-7B
Instruction-Oriented Preference Alignment for Enhancing Multi-Modal Comprehension Capability of MLLMs
ICCV 2025 · Table 1
Multimodal LLMs
33.49+ P2-DPO
Qwen2.5-VL-3B
P$^2$-DPO:Grounding Hallucination in Perceptual Processing via Calibration Direct Preference Optimization
ICLR 2026 · Table 2
Multimodal LLMs
43.31Lγ with RLAIF-V-Bias
LLaVA-v1.5-7B
Debiasing Multimodal Large Language Models via Noise-Aware Preference Optimization
CVPR 2025 · Table 1
Multimodal LLMs
53.28REVERSE (τ=0.0003)
LLaVA-v1.5 7B
Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling
NeurIPS 2025 · MMHal-Bench table
Multimodal LLMs
63.23AMP-MEG 13B
LLaVA-v1.5 13B
Automated Multi-level Preference for MLLMs
NeurIPS 2024 · Table: Comparison of conventio
Multimodal LLMs
73.16SCPO
LLaVA-v1.6-7B
Mitigating Visual Hallucinations via Semantic Curriculum Preference Optimization in MLLMs
ICML 2026 · Table 1
Multimodal LLMs
83.12LLaVA-1.5 + AdaViP
LLaVA-1.5 7B
AdViP: Aligning Multi-modal LLMs via Adaptive Vision-enhanced Preference Optimization
IJCV 2026 · Table 1
Multimodal LLMs
93.07OPA-DPO
LLaVA-Instruct-1.5-13B
Mitigating Hallucinations in Large Vision-Language Models via DPO: On-Policy Data Hold the Key
CVPR 2025 · Benchmark Evaluation Results t
Multimodal LLMs
103.06TPR-CL-7B
LLaVA-1.5-7B
Systematic Reward Gap Optimization for Mitigating VLM Hallucinations
NeurIPS 2025 · Table 1
Multimodal LLMs
13 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.