| One Model, Many Budgets: Elastic Latent Interfaces for Diffusion Transformers | ✓ Link | 90.45 | ELIT-Qwen-Image (Qwen-Image, 20B) | 2026-03-12 |
| HiDream-O1-Image: A Natively Unified Image Generative Foundation Model with Pixel-level Unified Transformer | ✓ Link | 90.30 | HiDream-O1-Image-Pro | 2026-05-11 |
| HiDream-O1-Image: A Natively Unified Image Generative Foundation Model with Pixel-level Unified Transformer | ✓ Link | 89.83 | HiDream-O1-Image | 2026-05-11 |
| Nucleus-Image: Sparse MoE for Image Generation | ✓ Link | 88.79 | Nucleus-Image | 2026-04-14 |
| Simplifying and Improving On-Policy Distillation for Few-Step Diffusion | | 88.7 | SMD (Qwen-Image, 4-NFE) | 2026-10-13 |
| LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes | ✓ Link | 88.55 | LLaDA-Image-Turbo | 2026-09-03 |
| Parallel Decoding Distillation for Fast Image and Video Generation | ✓ Link | 88.46 | PDD (Midpoint) | 2026-07-28 |
| Boogu-Image-0.1: Boosting Open-Source Unified Multimodal Understanding and Generation | ✓ Link | 88.35 | Boogu-Image-0.1-Turbo | 2026-07-14 |
| Qwen-Image Technical Report | ✓ Link | 88.32 | Qwen-Image | 2025-08-04 |
| BitDance: Scaling Autoregressive Generative Models with Binary Tokens | ✓ Link | 88.3 | BitDance-14B-64x | 2026-02-15 |
| Reconstruction Alignment Improves Unified Multimodal Models | ✓ Link | 88.15 | Harmon-1.5B-RecA-plus (extra data) | 2025-09-08 |
| Z-Image: An Efficient Image Generation Foundation Model with Single-Stream Diffusion Transformer | ✓ Link | 88.14 | Z-Image | 2025-11-27 |
| SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecture | ✓ Link | 88.14 | SenseNova-U1 (8BA3B) | 2026-05-12 |
| SenseNova-U1.5: Towards Native Unified Visual Intelligence | ✓ Link | 88.11 | SenseNova-U1.5 | 2026-09-10 |
| Read It Back: Pretrained MLLMs Are Zero-Shot Reward Models for Text-to-Image Generation | ✓ Link | 88.08 | SpectraReward (512px) | 2026-07-13 |
| Awaking Spatial Intelligence in Unified Multimodal Understanding and Generation | ✓ Link | 88.05 | JoyAI-Image | 2026-05-05 |
| Multi-Resolution Flow Matching: Training-Free Diffusion Acceleration via Staged Sampling | ✓ Link | 88.00 | MrFlow + Qwen-Image (20+1 steps) | 2026-07-02 |
| Intern Lumina U2: A Multi-Codebook Diffusion Large Language Model for Omni-Visual Understanding and Image Generation | ✓ Link | 87.99 | Intern Lumina U2 (NVIDIA) | 2026-10-14 |
| Pluggable Pruning with Contiguous Layer Distillation for Diffusion Transformers | ✓ Link | 87.9 | PPCL (Qwen-Image, 14B) | 2025-11-20 |
| DeepGen 1.0: A Lightweight Unified Multimodal Model for Advancing Image Generation and Editing | ✓ Link | 87.90 | DeepGen-1.0 (RL) | 2026-02-12 |
| AlphaGRPO: Unlocking Self-Reflective Multimodal Generation in UMMs via Decompositional Verifiable Reward | ✓ Link | 87.9 | AlphaGRPO + Self-Reflective Refinement | 2026-05-12 |
| APT: Feedback-Driven Agentic Post-Training for Text-to-Image Generation | ✓ Link | 87.89 | LongCat-Image (w/ APT) | 2026-10-13 |
| TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows | ✓ Link | 87.8 | Qwen-Image-TwinFlow (full, longer training) | 2025-12-03 |
| LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model | ✓ Link | 87.76 | LLaDA2.0-Uni (16B) | 2026-04-22 |
| X-Omni: Reinforcement Learning Makes Discrete Autoregressive Image Generative Models Great Again | ✓ Link | 87.65 | X-Omni | 2025-07-29 |
| An Empirical Study of Training Pixel-Space Text-to-Image Diffusion Models | | 87.6 | Ours (Z-Image pixel-space) (extra data) | 2026-08-17 |
| FLUX.2: Frontier Visual Intelligence | ✓ Link | 87.57 | FLUX.2 [Dev] | 2025-11-25 |
| GenRouter: Unified Workflow Routing for Agentic Image Generation | ✓ Link | 87.39 | GenCanvas + GenRouter (Qwen-Image) | 2026-08-17 |
| Refinement via Regeneration: Enlarging Modification Space Boosts Image Refinement in Unified Multimodal Models | ✓ Link | 87.21 | BAGEL-RvR | 2026-04-28 |
| SnapGen++: Unleashing Diffusion Transformers for Efficient High-Fidelity Image Generation on Edge Devices | | 87.2 | SnapGen++ (full, 1.6B) | 2026-01-13 |
| Trust Your Critic: Robust Reward Modeling and Reinforcement Learning for Faithful Image Editing and Generation | ✓ Link | 87.16 | FIRM-SD3.5 | 2026-03-12 |
| Learning Representations to Guide Pixel Diffusion Transformers | ✓ Link | 87.1 | RG-T2I-L/16 | 2026-10-09 |
| LLaDA-o: An Effective and Length-Adaptive Omni Diffusion Model | ✓ Link | 87.04 | LLaDA-o | 2026-03-01 |
| Nemotron-Labs-Diffusion-Image: Advancing Masked Discrete Diffusion for High-Resolution Image Synthesis | | 86.9 | Nemotron-Labs-Diffusion-Image (6M synthetic finetune) | 2026-06-29 |
| LongCat-Image Technical Report | ✓ Link | 86.8 | LongCat-Image | 2025-12-08 |
| UniCorn: Towards Self-Improving Unified Multimodal Models through Self-Generated Supervision | ✓ Link | 86.8 | UniCorn | 2026-01-06 |
| TUNA: Taming Unified Visual Representations for Native Unified Multimodal Models | ✓ Link | 86.76 | Tuna | 2025-12-01 |
| i1: A Simple and Fully Open Recipe for Strong Text-to-Image Models | ✓ Link | 86.73 | i1 3B | 2026-06-09 |
| Channel-wise Vector Quantization | ✓ Link | 86.72 | CAR-8B | 2026-05-25 |
| VQRAE: Representation Quantization Autoencoders for Multimodal Understanding, Generation and Reconstruction | | 86.67 | VQRAE | 2025-11-28 |
| Ovis-Image Technical Report | ✓ Link | 86.59 | Ovis-Image (2B+7B) | 2025-11-28 |
| Enhancing Alignment for Unified Multimodal Models via Semantically-Grounded Supervision | | 86.58 | SeGroS-Harmon-1.5B (extra data) | 2026-03-20 |
| Tuna-2: Pixel Embeddings Beat Vision Encoders for Multimodal Understanding and Generation | ✓ Link | 86.54 | Tuna-2 7B | 2026-04-27 |
| Iris-3B: Going Beyond the Latent with Pixel-Space Diffusion Training, Conversion and Fine-Tuning | ✓ Link | 86.52 | Iris-3B | 2026-10-07 |
| Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editing | ✓ Link | 86.49 | Mage-Flow | 2026-07-21 |
| UniSpace: Unified Visual Representation and Scalable Multimodal Modeling | ✓ Link | 86.49 | UniSpace | 2026-08-09 |
| Illuminating Unified Multimodal Model for Free-form Interleaved Text-Image Generation | ✓ Link | 86.38 | ILLUME-X | 2026-06-29 |
| HiRAE: Hierarchical Representation Autoencoding with Residual Budgets | | 86.35 | HiRAE-24 | 2026-09-29 |
| Self-Adversarial One Step Generation via Condition Shifting | ✓ Link | 86.17 | APEX 20B (LoRA&r=32) | 2026-04-14 |
| Steering Visual Generation in Unified Multimodal Models with Understanding Supervision | | 86.12 | BAGEL-7B (UNO) | 2026-05-07 |
| Echo-4o: Harnessing the Power of GPT-4o Synthetic Images for Improved Image Generation | ✓ Link | 86.07 | Echo-4o | 2025-08-13 |
| RubricRL: Simple Generalizable Rewards for Text-to-Image Generation | | 86.07 | RubricRL (Phi3-3.8B) (extra data) | 2025-11-25 |
| StageVAR: Stage-Aware Acceleration for Visual Autoregressive Models | ✓ Link | 86.05 | Infinity-8B + StageVAR | 2025-12-18 |
| DC-SAE: Deep Compression Semantic Autoencoder for Faster Diffusion Convergence | ✓ Link | 86.007 | 1.6B-DiT (extra data) | 2026-09-30 |
| Uniform Discrete Diffusion with Metric Path for Video Generation | ✓ Link | 86.0 | URSA-1.7B (1024x1024) | 2025-10-28 |
| L2P: Unlocking Latent Potential for Pixel Generation | ✓ Link | 86.00 | L2P | 2026-05-12 |
| UniCom: Unified Multimodal Modeling via Compressed Continuous Semantic Representations | | 85.92 | UniCom | 2026-03-11 |
| MiniT2I: A Minimalist Baseline for Text-to-Image Generation | ✓ Link | 85.9 | MiniT2I-L/16 | 2026-06-14 |
| High-Fidelity Two-Step Image Generation via Teacher-Aligned End-to-End Distillation | | 85.86 | Z-Image Turbo++ | 2026-06-10 |
| Decoupled DMD: CFG Augmentation as the Spear, Distribution Matching as the Shield | ✓ Link | 85.85 | Lumina-Image-2.0 + Decoupled DMD (Decoupled-Hybrid) | 2025-11-27 |
| DreamLite: A Lightweight On-Device Unified Model for Image Generation and Editing | ✓ Link | 85.8 | DreamLite | 2026-03-30 |
| SVG-T2I: Scaling Up Text-to-Image Latent Diffusion Model Without Variational Autoencoder | ✓ Link | 85.78 | SVG-T2I | 2025-12-12 |
| Auto-Rubric as Reward: From Implicit Preferences to Explicit Multimodal Generative Criteria | ✓ Link | 85.76 | FLUX.1-dev + ARR-RPO (Gemini 3.1 Pro judge) | 2026-05-08 |
| LatentUMM: Dual Latent Alignment for Unified Multimodal Models | ✓ Link | 85.74 | Harmon + LatentUMM | 2026-05-18 |
| EMMA: Efficient Multimodal Understanding, Generation, and Editing with a Unified Architecture | ✓ Link | 85.63 | EMMA | 2025-12-04 |
| Skywork UniPic: Unified Autoregressive Modeling for Visual Understanding and Generation | ✓ Link | 85.50 | Skywork UniPic | 2025-08-05 |
| NextStep-1: Toward Autoregressive Image Generation with Continuous Tokens at Scale | ✓ Link | 85.28 | NextStep-1 | 2025-08-14 |
| Continuous-Time Distribution Matching for Few-Step Diffusion Distillation | ✓ Link | 85.26 | CDM (SD3-Medium) | 2026-05-07 |
| Continuous Adversarial Flow Models | | 85.21 | CAFM | 2026-04-13 |
| InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editing | ✓ Link | 85.18 | InternVL-U (4B) | 2026-03-10 |
| Chimera: Designing and Chinchilla-Scaling Hybrid Visual Diffusion Transformers | | 85.12 | Chimera | 2026-07-30 |
| RepFusion: Leveraging Multimodal Priors for Denoising in Representation Space | | 85.11 | RepFusion-SFT w/ Diffusion Decoder | 2026-06-12 |
| Enhancing Spatial Understanding in Image Generation via Reward Modeling | ✓ Link | 85.03 | FLUX.1-dev (GRPO + SpatialScore) | 2026-02-27 |
| STARFlow2: Bridging Language Models and Normalizing Flows for Unified Multimodal Generation | | 84.94 | STARFlow2 (10.6B) | 2026-05-08 |
| Lance: Unified Multimodal Modeling by Multi-Task Synergy | ✓ Link | 84.67 | Lance (3B) | 2026-05-18 |
| GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling | | 84.65 | GenFirst EiT (1.3B, Qwen3-1.7B text encoder) | 2026-08-29 |
| Linearizing Vision Transformer with Test-Time Training | ✓ Link | 84.43 | SD3.5-T 5 | 2026-05-28 |
| Both Semantics and Reconstruction Matter: Making Representation Encoders Ready for Text-to-Image Generation and Editing | | 84.25 | PS-VAE (32c) | 2025-12-19 |
| Representation Forcing for Bottleneck-Free Unified Multimodal Models | | 84.15 | RF-Pixel (Pixel+RF) | 2026-05-29 |
| InsightTok: Improving Text and Face Fidelity in Discrete Tokenization for Autoregressive Image Generation | ✓ Link | 84.11 | InsightAR | 2026-05-14 |
| Energy-Guided Flow Matching | ✓ Link | 83.9 | EG-FM-T2I | 2026-08-06 |
| PixelDiT: Pixel Diffusion Transformers for Image Generation | ✓ Link | 83.7 | PixelDiT-T2I | 2025-11-25 |
| Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation | ✓ Link | 83.48 | Cheers (1.5B) | 2026-03-13 |
| ParaUni: Enhance Generation in Unified Multimodal Model with Reinforcement-driven Hierarchical Parallel Information Interaction | ✓ Link | 83.45 | ParaUni | 2025-12-05 |
| Multimodal Flow: Unified Flow Modeling of Language and Vision in Embedding Spaces | ✓ Link | 83.44 | MF-1 | 2026-09-30 |
| Normalizing Trajectory Models | | 83.38 | NTM (finetune, 512×512) | 2026-05-08 |
| Adversarial Training for Pixel Diffusion | | 83.3 | DeCo +GAN | 2026-09-29 |
| Uncovering Understanding-Generation Synergy in Native Unified Multimodal Models: From Representation, Task to System | | 83.09 | Native UMM Synergy: Modality-decoupled MoT U+G | 2026-09-01 |
| DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation | ✓ Link | 81.4 | DeCo-XXL/16 | 2025-11-24 |
| Extending One-Step Image Generation from Class Labels to Text via Discriminative Text Representation | ✓ Link | 81.2 | EMF (3B) | 2026-04-20 |
| On the Diffusibility of High-Dimensional Latents | | 81.15 | On the Diffusibility of High-Dimensional Latents (Ours, 512) | 2026-09-23 |
| PixNerd: Pixel Neural Field Diffusion | ✓ Link | 80.9 | PixNerd-XXL/16 | 2025-07-31 |
| ELLA: Equip Diffusion Models with LLM for Enhanced Semantic Alignment | ✓ Link | 80.23 | ELLA SDXL | 2024-03-08 |
| Few-Step Distillation for Text-to-Image Generation: A Practical Guide | ✓ Link | 80.03 | MeanFlow (4 NFE) (extra data) | 2025-12-15 |
| PixelDense: Dense Prediction as Representation Alignment for Pixel Diffusion | ✓ Link | 78.9 | PixelDense | 2026-09-30 |
| Compositional Text-to-Image Generation Via Region-aware Bimodal Direct Preference Optimization | ✓ Link | 78.84 | SDXL-BiDPO | 2026-05-27 |
| Visual Generation Tuning | ✓ Link | 78.73 | VGT (Qwen2.5-VL) | 2025-11-28 |
| PixelFlow: Pixel-Space Generative Models with Flow | ✓ Link | 77.93 | PixelFlow | 2025-04-10 |
| Text-to-Image Generative Modeling Without Language Pretraining | | 77.9 | Halcyon-1.60B | 2026-10-19 |
| RoAR-VAE: Improving Latent Diffusability through Robust Prediction | ✓ Link | 77.56 | RoAR-VAE | 2026-10-19 |
| Scaling Text-to-Image Diffusion Transformers with Representation Autoencoders | ✓ Link | 76.9 | RAE (SigLIP-2 ViT-So) | 2026-01-22 |
| Cross-Space Distillation: Teaching One-Step Students with Modern Diffusion Teachers | | 68.52 | SD 2.1 + Bridge (SD 3.5 Medium) | 2026-06-30 |