DeepGen 1.0: A Lightweight Unified Multimodal Model for Advancing Image Generation and Editing

2026-02-12

AI Summary

This work proposes DeepGen 1.0, a lightweight 5-billion-parameter unified model for multimodal image generation and editing that significantly reduces the computational cost typically associated with large-scale models exceeding 10 billion parameters. DeepGen 1.0 introduces a novel Stacked Channel Bridging mechanism to deeply integrate multi-level features from vision-language models with learnable “thought tokens,” enabling fine-grained semantic control. The model is trained via a three-stage strategy—alignment pretraining, joint fine-tuning, and MR-GRPO reinforcement learning—achieving strong performance despite using only 50 million training samples. On the WISE benchmark, DeepGen 1.0 outperforms HunyuanImage (80B parameters) by 28%, and surpasses Qwen-Image-Edit (27B parameters) by 37% on UniREditBench. The code, model weights, and dataset are publicly released.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Current unified multimodal models for image generation and editing typically rely on massive parameter scales (e.g.,>10B), entailing prohibitive training costs and deployment footprints. In this work, we present DeepGen 1.0, a lightweight 5B unified model that achieves comprehensive capabilities competitive with or surpassing much larger counterparts. To overcome the limitations of compact models in semantic understanding and fine-grained control, we introduce Stacked Channel Bridging (SCB), a deep alignment framework that extracts hierarchical features from multiple VLM layers and fuses them with learnable'think tokens'to provide the generative backbone with structured, reasoning-rich guidance. We further design a data-centric training strategy spanning three progressive stages: (1) Alignment Pre-training on large-scale image-text pairs and editing triplets to synchronize VLM and DiT representations, (2) Joint Supervised Fine-tuning on a high-quality mixture of generation, editing, and reasoning tasks to foster omni-capabilities, and (3) Reinforcement Learning with MR-GRPO, which leverages a mixture of reward functions and supervision signals, resulting in substantial gains in generation quality and alignment with human preferences, while maintaining stable training progress and avoiding visual artifacts. Despite being trained on only ~50M samples, DeepGen 1.0 achieves leading performance across diverse benchmarks, surpassing the 80B HunyuanImage by 28% on WISE and the 27B Qwen-Image-Edit by 37% on UniREditBench. By open-sourcing our training code, weights, and datasets, we provide an efficient, high-performance alternative to democratize unified multimodal research.

Problem

• multimodal models
• image generation
• image editing
• model scale
• training cost

Innovation

• Stacked Channel Bridging
• lightweight unified multimodal model
• MR-GRPO reinforcement learning
• think tokens
• data-centric training strategy