| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 85.0 | Qwen3.5-397B-A17B | 2026-02-16 |
| Kimi K2.5: Visual Agentic Intelligence | ✓ Link | 84.3 | Kimi K2.5 | 2026-02-02 |
| Introducing GPT-5 | | 84.2 | GPT-5 | 2025-08-07 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 83.9 | Qwen3.5-122B-A10B | 2026-02-16 |
| Introducing OpenAI o3 and o4-mini | ✓ Link | 82.9 | OpenAI o3 | 2025-04-16 |
| Qwen3.6 | ✓ Link | 82.9 | Qwen3.6-27B | 2026-04-21 |
| HopChain: Multi-Hop Data Synthesis for Generalizable Vision-Language Reasoning | | 82.89 | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | 2026-03-17 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 82.3 | Qwen3.5-27B | 2026-02-16 |
| Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities | | 82.0 | Gemini 2.5 Pro | 2025-07-07 |
| Qwen3.6 | ✓ Link | 81.7 | Qwen3.6-35B-A3B | 2026-04-21 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 81.4 | Qwen3.5-35B-A3B | 2026-02-16 |
| Qwen3-VL Technical Report | ✓ Link | 80.6 | Qwen3-VL-235B-A22B-Thinking | 2025-11-26 |
| SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecture | ✓ Link | 80.55 | SenseNova-U1-A3B-MoT | 2026-05-12 |
| STEP3-VL-10B Technical Report | ✓ Link | 80.11 | Step3-VL-10B (PaCoRe) | 2026-01-14 |
| Qwen3.5-Omni Technical Report | | 80.1 | Qwen3.5-Omni-Plus | 2026-04-21 |
| Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities | | 79.7 | Gemini 2.5 Flash | 2025-07-07 |
| EXAONE 4.5 Technical Report | ✓ Link | 78.7 | EXAONE 4.5 (Reasoning) | 2026-04-09 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 78.4 | Qwen3.5-9B | 2026-02-16 |
| Introducing OpenAI o1 | | 78.2 | OpenAI o1 | 2024-09-12 |
| Claude Opus 4.1 | | 77.1 | Claude Opus 4.1 | 2025-08-05 |
| GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning | ✓ Link | 76.0 | GLM-4.6V | 2025-07-01 |
| Evidence-RL: Towards Evidence-intensive Visual Reasoning | | 75.67 | Evidence-RL (Qwen3.5-9B) | 2026-08-08 |
| Kwai Keye-VL-2.0 Technical Report | ✓ Link | 75.3 | Keye-VL-2.0-30B-A3B | 2026-06-09 |
| SpatialASM: Internalizing Active Mentalization for Spatial Reasoning | ✓ Link | 75.3 | SpatialASM-Qwen3.5-9B | 2026-10-09 |
| Command A+ | | 75.1 | Command A+ | 2026-05-20 |
| Qwen3-Omni Technical Report | ✓ Link | 75.0 | Qwen3-Omni-Flash-Thinking | 2025-09-22 |
| Masking Teacher and Reinforcing Student for Distilling Vision-Language Models | | 74.0 | InternVL3 - Masters | 2025-12-23 |
| SenseNova-U1.5: Towards Native Unified Visual Intelligence | ✓ Link | 73.86 | SenseNova-U1.5 8B-Think | 2026-09-10 |
| Introducing Claude 4 | | 73.7 | Claude Opus 4 | 2025-05-22 |
| The Llama 4 herd: The beginning of a new era of natively multimodal AI innovation | ✓ Link | 73.4 | Llama 4 Maverick | 2025-04-05 |
| TLive-Omni: An Omni-Modal Understanding Model for E-Commerce Live Streaming | ✓ Link | 73.4 | TLive-Omni-9B | 2026-08-21 |
| Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving | ✓ Link | 72.67 | Qwen-Drive-1.0-SFT | 2026-08-31 |
| Introducing Claude 4 | | 72.6 | Claude Sonnet 4 | 2025-05-22 |
| InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models | ✓ Link | 72.2 | InternVL3-78B | 2025-04-14 |
| Dynamic Important Example Mining for Reinforcement Finetuning | ✓ Link | 71.9 | Qwen2.5-VL-32B + DIEM | 2026-08-29 |
| Claude 3.7 Sonnet and Claude Code | | 71.8 | Claude 3.7 Sonnet | 2025-02-24 |
| OpenVLThinkerV2: A Generalist Multimodal Reasoning Model for Multi-domain Visual Tasks | ✓ Link | 71.6 | OpenVLThinkerV2 (Ours) | 2026-04-09 |
| SPOT-E: Test-Time Entropy Shaping with Visual Spotlights for Frozen VLMs | ✓ Link | 71.5 | SPOT-E (Gemini-2.5-Flash) | 2026-06-18 |
| Kwai Keye-VL Technical Report | ✓ Link | 71.4 | Keye-VL-8B | 2025-07-02 |
| Kwai Keye-VL 1.5 Technical Report | ✓ Link | 71.4 | Keye-VL-1.5-8B | 2025-09-01 |
| Ovis2.5 Technical Report | ✓ Link | 71.2 | Ovis2.5-9B | 2025-08-15 |
| Nemotron 3 Nano Omni: Efficient and Open Multimodal Intelligence | ✓ Link | 70.8 | Nemotron 3 Nano Omni (Reasoning) | 2026-04-27 |
| Odysseus: Scaling VLMs to 100+ Turn Decision-Making in Games via Reinforcement Learning | | 70.77 | Odysseus (RL on SFT) | 2026-05-01 |
| Introducing Gemini 2.0: our new AI model for the agentic era | | 70.7 | Gemini 2.0 Flash Experimental | 2024-12-11 |
| OneThinker: All-in-one Reasoning Model for Image and Video | ✓ Link | 70.6 | OneThinker-8B | 2025-12-02 |
| LongCat-Next: Lexicalizing Modalities as Discrete Tokens | ✓ Link | 70.6 | LongCat-Next (A3B, 68.5B) | 2026-03-29 |
| V-Rubrics: Visual Faithfulness via Rubric-Based Reinforcement Learning | ✓ Link | 70.56 | V-Rubrics (GRPO w/ rubrics) (extra data) | 2026-08-26 |
| Qwen2.5-VL Technical Report | ✓ Link | 70.2 | Qwen2.5-VL-72B | 2025-02-19 |
| Context-Aware RL for Agentic and Multimodal LLMs | ✓ Link | 70.1 | ContextRL (trained from Qwen3-VL-8B) | 2026-06-15 |
| Beyond Simply Environment Scaling: Designing Effective Environment Distributions for Multimodal Agent Learning | ✓ Link | 70.1 | AES+HDC (Qwen3-VL-8B-Instruct) | 2026-08-06 |
| ERNIE 4.5 Technical Report | ✓ Link | 70.0 | ERNIE-4.5-VL-424B-A47B | 2025-06-30 |
| The Llama 4 herd: The beginning of a new era of natively multimodal AI innovation | ✓ Link | 69.4 | Llama 4 Scout | 2025-04-05 |
| Hello GPT-4o | | 69.1 | GPT-4o | 2024-05-13 |
| MiniMax-01: Scaling Foundation Models with Lightning Attention | ✓ Link | 68.5 | MiniMax-VL-01 | 2025-01-14 |
| Can We Trust the Teacher? Decoupled Credit Direction-Magnitude for Self-Distillation | | 68.44 | DCSD (Qwen3-VL-8B-Instruct) | 2026-09-28 |
| Introducing Claude 3.5 Sonnet | | 68.3 | Claude 3.5 Sonnet | 2024-06-20 |
| EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience | ✓ Link | 68.11 | EvoCUA-32B | 2026-01-22 |
| From Pixels to Words -- Towards Native One-Vision Models at Scale | ✓ Link | 68.1 | NEO-ov 9B | 2026-05-27 |
| NVIDIA Nemotron Nano V2 VL | ✓ Link | 67.8 | Nemotron Nano V2 VL (Reasoning on) (extra data) | 2025-11-06 |
| Co-Evolving Policy Distillation | | 67.78 | CoPD | 2026-04-29 |
| MiniCPM-V 4.5: Cooking Efficient MLLMs via Architecture, Data, and Training Recipe | ✓ Link | 67.7 | MiniCPM-V 4.5 | 2025-09-16 |
| STARFlow2: Bridging Language Models and Normalizing Flows for Unified Multimodal Generation | | 67.7 | STARFlow2 (10.6B) | 2026-05-08 |
| iVGR: Internalizing Visually Grounded Reasoning for MLLMs with Reinforcement Learning | ✓ Link | 67.7 | iVGR-Qwen3-VL-32B (ours) | 2026-05-29 |
| MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction | ✓ Link | 67.6 | MiniCPM-o 4.5-Instruct | 2026-04-30 |
| Wake up for Touch! Mask-isolated Tactile Alignment Learning in MLLMs | ✓ Link | 67.4 | Splash (3B) | 2026-07-01 |
| Self-Distilled RLVR | | 67.22 | RLSD (Qwen3-VL-8B-Instruct) | 2026-04-03 |
| FocusVTC: Efficient and High-Performance Visual Text Compression with Adaptive Resolution | ✓ Link | 66.73 | FocusVTC | 2026-09-29 |
| Beyond SFT-to-RL: Pre-alignment via Black-Box On-Policy Distillation for Multimodal RL | ✓ Link | 66.6 | PRISM+GRPO (Qwen3-VL-8B) | 2026-05-01 |
| EviRover: Reinforcing Agentic Perception Beyond a Glance | ✓ Link | 66.2 | EviRover | 2026-09-30 |
| Introducing Command A Vision: Multimodal AI built for business | | 65.3 | Command A Vision | 2025-07-31 |
| Insight-V++: Towards Advanced Long-Chain Visual Reasoning with Multimodal Large Language Models | | 64.8 | Insight-V++ | 2026-03-18 |
| The Llama 3 Herd of Models | ✓ Link | 64.5 | Llama 3-V (405B) | 2024-07-31 |
| Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution | ✓ Link | 64.5 | Qwen2-VL-72B | 2024-09-18 |
| Kimi-VL Technical Report | ✓ Link | 64.0 | Kimi-VL-A3B-Thinking-2506 | 2025-04-10 |
| Learning Adaptive Reasoning Paths for Efficient Visual Reasoning | ✓ Link | 64.0 | AVR-8B | 2026-04-16 |
| DeepAlign: Mitigating Modality Conflict through Modality-Specific Alignment | | 63.7 | DeepAlign (InternVL3-8B) | 2026-06-03 |
| Unlocking Token Rewards via Training-Free Reward Attribution | | 63.4 | Qwen2.5-VL-7B-Instruct + GRPO (P2T token reward) (extra data) | 2026-06-03 |
| LLaVA-UHD v4: What Makes Efficient Visual Encoding in MLLMs? | ✓ Link | 62.3 | LLaVA-UHD v4 | 2026-05-09 |
| Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context | ✓ Link | 62.2 | Gemini 1.5 Pro | 2024-03-08 |
| LASER: A Corrective Lens for LVLMs via Visual Attention Preservation and Sink Suppression | ✓ Link | 62.1 | LASER | 2026-07-02 |
| Infinity-Parser2-Pro | | 61.89 | Infinity-Parser2-Pro | 2026-05-11 |
| VQRAE: Representation Quantization Autoencoders for Multimodal Understanding, Generation and Reconstruction | | 61.6 | VQRAE (InternViT-300M, Qwen2.5-7B, 448px) | 2025-11-28 |
| Macaron-V1: Towards Open Continual Learning with Self-Improvement and Mixture-of-LoRA | ✓ Link | 61.22 | Macaron-V1-Tall | 2026-08-10 |
| TUNA: Taming Unified Visual Representations for Native Unified Multimodal Models | ✓ Link | 61.2 | Tuna | 2025-12-01 |
| Learning from Primitive: Probing Visual Reasoning of LVLMs via Counting | | 60.74 | Qwen3-VL-8B (Learning from Primitive) (extra data) | 2026-09-08 |
| Llama 3.2: Revolutionizing edge AI and vision with open, customizable models | ✓ Link | 60.3 | Llama 3.2 90B Vision Instruct | 2024-09-25 |
| Ministral 3 | | 59.9 | Ministral 3 (14B) | 2026-01-13 |
| Gemini: A Family of Highly Capable Multimodal Models | ✓ Link | 59.4 | Gemini Ultra | 2023-12-19 |
| Introducing the next generation of Claude | | 59.4 | Claude 3 Opus | 2024-03-04 |
| GPT-4o mini: advancing cost-efficient intelligence | ✓ Link | 59.4 | GPT-4o mini | 2024-07-18 |
| CodeV: Code with Images for Faithful Visual Reasoning via Tool-Aware Policy Optimization | ✓ Link | 59.3 | CodeV-7B-RL | 2025-11-24 |
| PointThinker: Point-Incentivized Parallel Thinking for Multimodal Large Language Model | | 59.3 | PointThinker-7B (extra data) | 2026-06-03 |
| Qwen2.5-Omni Technical Report | ✓ Link | 59.2 | Qwen2.5-Omni-7B | 2025-03-26 |
| Lumina-DiMOO: An Omni Diffusion Large Language Model for Multi-Modal Generation and Understanding | ✓ Link | 58.6 | Lumina-DiMOO | 2025-10-07 |
| AnE: Pushing the Reasoning Frontier of Multimodal LLMs via Anchor Evolution | ✓ Link | 58.1 | AnE-3rd | 2026-05-25 |
| ARM-Thinker: Reinforcing Multimodal Generative Reward Models with Agentic Tool Use and Visual Reasoning | ✓ Link | 57.2 | ARM-Thinker-7B | 2025-12-04 |
| Kimi-VL Technical Report | ✓ Link | 57.0 | Kimi-VL-A3B-Instruct | 2025-04-10 |
| Generation as Auxiliary Supervision: Enhancing Visual Understanding at Zero Inference Overhead via Decoupled Embedding Prediction | | 57.0 | GAS (4B) | 2026-08-12 |
| LLaVA-OneVision: Easy Visual Task Transfer | ✓ Link | 56.8 | LLaVA-OneVision-72B | 2024-08-06 |
| LLaDA-UI: Bringing Block-wise Diffusion to Vision-Language GUI Agents | ✓ Link | 56.2 | LLaDA-UI (pre-training) | 2026-09-09 |
| Oryx MLLM: On-Demand Spatial-Temporal Understanding at Arbitrary Resolution | ✓ Link | 56.1 | Oryx-1.5 32B | 2024-09-19 |
| SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem | ✓ Link | 55.5 | SpatialBlock-7B-reason | 2026-09-07 |
| Intern Lumina U2: A Multi-Codebook Diffusion Large Language Model for Omni-Visual Understanding and Image Generation | ✓ Link | 55.5 | Intern Lumina U2 (NVIDIA) | 2026-10-14 |
| Emerging Properties in Unified Multimodal Pretraining | ✓ Link | 55.3 | BAGEL (7B MoT) | 2025-05-20 |
| Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs | ✓ Link | 55.1 | Phi-4-Multimodal | 2025-02-24 |
| Aria: An Open Multimodal Native Mixture-of-Experts Model | ✓ Link | 54.9 | Aria | 2024-10-08 |
| Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Models | ✓ Link | 54.8 | Embodied-R1.5 | 2026-06-09 |
| InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editing | ✓ Link | 54.7 | InternVL-U (4B) | 2026-03-10 |
| From Pixels to Words -- Towards Native Vision-Language Primitives at Scale | ✓ Link | 54.6 | NEO-9B | 2025-10-16 |
| LatentUM: Unleashing the Potential of Interleaved Cross-Modal Reasoning via a Latent-Space Unified Model | ✓ Link | 54.6 | LatentUM (Base) | 2026-04-02 |
| Uni-Edit: Intelligent Editing Is A General Task For Unified Model Tuning | ✓ Link | 54.2 | BAGEL-Uni-Edit (stage 2) | 2026-05-20 |
| Representation Forcing for Bottleneck-Free Unified Multimodal Models | | 54.2 | Pixel+RF | 2026-05-29 |
| UniPath: Adaptive Coordination of Understanding and Generation for Unified Multimodal Reasoning | | 54.11 | UniPath (BAGEL backbone) | 2026-05-12 |
| CEPO: RLVR Self-Distillation using Contrastive Evidence Policy Optimization | ✓ Link | 54.11 | Qwen3-VL-4B-Instruct with CEPO | 2026-05-19 |
| Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution | ✓ Link | 54.1 | Qwen2-VL-7B | 2024-09-18 |
| Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Models | ✓ Link | 54.1 | Molmo 72B | 2024-09-25 |
| VideoAuto-R1: Video Auto Reasoning via Thinking Once, Answering Twice | ✓ Link | 53.8 | VideoAuto-R1 (Qwen2.5-VL-7B) | 2026-01-08 |
| Decouple to Generalize: Context-First Self-Evolving Learning for Data-Scarce Vision-Language Reasoning | ✓ Link | 53.6 | DoGe-7B (Iter3) | 2025-12-07 |
| Reinforcement Learning for Multimodal Diffusion Language Models via Bidimensional Trajectory and Thought Optimization | | 53.4 | LLaDA-V + Bi-VRL (ours) | 2026-09-08 |
| LatentUMM: Dual Latent Alignment for Unified Multimodal Models | ✓ Link | 53.2 | Bagel (14B) + LatentUMM | 2026-05-18 |
| Introducing the next generation of Claude | | 53.1 | Claude 3 Sonnet | 2024-03-04 |
| Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding | ✓ Link | 53.0 | Molmo2-8B | 2026-01-15 |
| Paying More Attention to Visual Tokens in Self-Evolving Large Multimodal Models | ✓ Link | 52.69 | VISE (Qwen3-VL-8B) | 2026-06-25 |
| UniDDT: Unifying Multimodal Understanding and Generation with Decoupled Diffusion Transformer | | 52.6 | VLM-UniDDT | 2026-06-15 |
| VisionFoundry: Teaching VLMs Visual Perception with Synthetic Images | ✓ Link | 52.3 | MiMo-VL-7B-SFT + VisionFoundry-10K | 2026-04-10 |
| Dream-VL & Dream-VLA: Open Vision-Language and Vision-Language-Action Models with Diffusion Language Model Backbone | ✓ Link | 52.2 | Dream-VL | 2025-12-27 |
| Pixtral 12B | ✓ Link | 52.0 | Pixtral 12B | 2024-10-09 |
| AutoV: Loss-Oriented Ranking for Visual Prompt Retrieval in LVLMs | | 52.0 | AutoV + Qwen2.5-VL (7B) | 2025-06-19 |
| DICS: Exploring Data Intrinsic Consistency for Visual Instruction Selection | ✓ Link | 52 | DICS (Qwen2-VL-7B, LLaVA-1.5-7B selector) | 2026-08-31 |
| Dynin-Omni: Omnimodal Unified Large Diffusion Language Model | ✓ Link | 51.6 | Dynin-Omni (8B) | 2026-03-09 |
| Harnessing Textual Refusal Directions for Multimodal Safety | ✓ Link | 51.4 | MARS (InternVL 3.5 8B) | 2026-10-16 |
| LoMo: Local Modality Substitution for Deeper Vision-Language Fusion | ✓ Link | 51.22 | LoMo (LLaVA-OneVision-1.5-8B) | 2026-05-28 |
| LLaVA-NeXT: Improved reasoning, OCR, and world knowledge | ✓ Link | 51.1 | LLaVA-NeXT-34B | 2024-01-30 |
| MODUS: Decoder-Only Any-to-Any Modeling of Diverse Modalities | ✓ Link | 51.1 | MODUS | 2026-07-06 |
| ROSE: Rotate Your Large Language Model to See | | 50.8 | ROSE-7B (extra data) | 2026-06-03 |
| BLIP3-o: A Family of Fully Open Unified Multimodal Models-Architecture, Training and Dataset | ✓ Link | 50.6 | BLIP3-o (8B) | 2025-05-14 |
| Introducing the next generation of Claude | | 50.2 | Claude 3 Haiku | 2024-03-04 |
| NaVLM-PVC: Progressive Visual Compression for Efficient Native-Resolution Encoding in MLLMs | | 50.2 | NaVLM-PVC | 2026-09-08 |
| LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model | ✓ Link | 50.1 | LLaDA2.0-Uni (16B) | 2026-04-22 |
| DiffusionVL: Translating Any Autoregressive Models into Diffusion Vision Language Models | ✓ Link | 49.3 | DiffusionVL | 2025-12-17 |
| VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding | ✓ Link | 48.8 | VideoLLaMA3 7B | 2025-01-22 |
| LLaDA-V: Large Language Diffusion Models with Visual Instruction Tuning | ✓ Link | 48.6 | LLaDA-V | 2025-05-22 |
| T5Gemma 2: Seeing, Reading, and Understanding Longer | | 47.8 | T5Gemma 2 4B-4B (post-trained) | 2025-12-16 |
| Information-Regularized Attention for Visual-Centric Reasoning | | 47.6 | InternVL2.5-IRA | 2026-07-01 |
| PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models | ✓ Link | 47.2 | PerceptionDLM-Base (8B) | 2026-06-17 |
| Program-Verified Self-Evolution for Vision-Language Models | ✓ Link | 46.78 | VQS-2B (Qwen3-VL-2B-Instruct) | 2026-09-27 |
| Ovis: Structural Embedding Alignment for Multimodal Large Language Model | ✓ Link | 46.7 | Ovis-Qwen1.5-14B | 2024-05-31 |
| Building and better understanding vision-language models: insights and future directions | ✓ Link | 46.6 | Idefics3-8B | 2024-08-22 |
| PerceptionLM: Open-Access Data and Models for Detailed Visual Understanding | ✓ Link | 46.1 | PLM-8B | 2025-04-17 |
| ZAYA1-VL-8B Technical Report | ✓ Link | 46.0 | ZAYA1-VL-8B | 2026-05-08 |
| MiniCPM-V: A GPT-4V Level MLLM on Your Phone | ✓ Link | 45.8 | MiniCPM-Llama3-V 2.5 | 2024-08-03 |
| FastVLM: Efficient Vision Encoding for Vision Language Models | ✓ Link | 45.4 | FastVLM-7B | 2024-12-17 |
| LFM2 Technical Report | ✓ Link | 45.33 | LFM2-VL-3B | 2025-11-28 |
| VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding | ✓ Link | 45.3 | VideoLLaMA3 2B | 2025-01-22 |
| Differences That Matter: Auditing Models for Capability Gap Discovery and Rectification | | 45.2 | Gemma3 + AuditDM | 2025-12-18 |
| LLaDA-o: An Effective and Length-Adaptive Omni Diffusion Model | ✓ Link | 44.9 | LLaDA-o | 2026-03-01 |
| Rethinking Projector Training in Multimodal LLMs | ✓ Link | 44.8 | PORTAL (Qwen 2.5 7B + CLIP) | 2026-10-13 |
| HyperVL: An Efficient and Dynamic Multimodal Large Language Model for Edge Devices | | 44.6 | HyperVL ViTL | 2025-12-16 |
| Unified Multimodal Autoregressive Modeling with Shared Context-Visual Tokenizer is Key to Unification | ✓ Link | 44.3 | UniAR | 2026-06-16 |
| InfiniteVL: Synergizing Linear and Sparse Attention for Highly-Efficient, Unlimited-Input Vision-Language Models | ✓ Link | 44.0 | InfiniteVL-Base | 2025-12-09 |
| Sparse-LaViDa: Sparse Multimodal Discrete Diffusion Language Models | | 43.6 | Sparse-LaViDa | 2025-12-16 |
| What matters when building vision-language models? | ✓ Link | 43.5 | Idefics2 (w/o image split) | 2024-05-03 |
| Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone | ✓ Link | 43.0 | Phi-3.5-Vision | 2024-04-22 |
| Uncovering Understanding-Generation Synergy in Native Unified Multimodal Models: From Representation, Task to System | | 42.89 | Native UMM Synergy: Modality-decoupled MoT U | 2026-09-01 |
| Decentralized Instruction Tuning: Conflict-Aware Splitting and Weight Merging | ✓ Link | 42.7 | MERIT (Proposed, 3D split, 8 groups) | 2026-06-01 |
| AIA: Rethinking Architecture Decoupling Strategy In Unified Multimodal Model | ✓ Link | 42.1 | Janus-Pro + AIA | 2025-11-27 |
| SmolVLM: Redefining small and efficient multimodal models | ✓ Link | 42.0 | SmolVLM 2.2B | 2025-04-07 |
| PixelUMM: Encoder-Free Unified Image and Video Understanding and Generation | ✓ Link | 41.67 | PixelUMM 8B MoT | 2026-09-29 |
| Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution | ✓ Link | 41.1 | Qwen2-VL-2B | 2024-09-18 |
| Scaling Text-to-Image Diffusion Transformers with Representation Autoencoders | ✓ Link | 41.1 | RAE (SigLIP-2) | 2026-01-22 |
| Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling | ✓ Link | 41.0 | Janus-Pro-7B | 2025-01-29 |
| Introducing LFM2.5: The Next Generation of On-Device AI | | 40.56 | LFM2.5-VL-1.6B | 2026-01-05 |
| Preventing Expert Collapse in MoE-dVLMs via Modality-Wise Norm Alignment | | 40.3 | TowerAlign | 2026-09-08 |
| ARM: An AutoRegressive Large Multimodal Model with Unified Discrete Representations | ✓ Link | 40.2 | ARM-7B | 2026-06-09 |
| ESC: Emotional Self-Correction for Reliable Vision-Language Models | ✓ Link | 40.02 | ESC + Qwen2-VL-7B | 2026-07-01 |
| Token-level Response-visual Attention Guidance for Multimodal LLMs Knowledge Distillation | ✓ Link | 38.7 | TRAG (Qwen2.5-1.5B) | 2026-07-01 |
| RubiCap: Rubric-Guided Reinforcement Learning for Dense Image Captioning | | 37.0 | RubiCap-7B (PixMoCap) | 2026-03-10 |
| DeepSeek-VL: Towards Real-World Vision-Language Understanding | ✓ Link | 36.6 | DeepSeek-VL 7B Chat | 2024-03-08 |
| Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation | ✓ Link | 36.0 | Cheers (1.5B) | 2026-03-13 |
| Switch-KD: Visual-Switch Knowledge Distillation for Vision-Language Models | ✓ Link | 34.9 | Switch-KD 1.5B | 2026-04-16 |
| Mobile-O: Unified Multimodal Understanding and Generation on Mobile Device | ✓ Link | 34.6 | Mobile-O-0.5B | 2026-02-23 |
| Modality Gap-Driven Subspace Alignment Training Paradigm For Multimodal Large Language Models | ✓ Link | 33.49 | ReVision-2M | 2026-02-02 |
| Mapping the Concept Landscape: Structural Perception of Global Distributions for Transparent Data Pruning | | 33.4 | MCL (Ours) | 2026-08-24 |
| LFM2.5-VL-450M: Structured Visual Intelligence, Edge to Cloud | | 32.67 | LFM2.5-VL-450M | 2026-04-08 |
| Emu3: Next-Token Prediction is All You Need | ✓ Link | 31.6 | Emu3 | 2024-09-27 |
| Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generation | ✓ Link | 30.5 | Janus | 2024-10-17 |