Unified Thinker: A General Reasoning Modular Core for Image Generation

TL;DR

Unified Thinker introduces a modular reasoning core for image generation that enhances logic-based instruction following by decoupling reasoning from image synthesis, enabling flexible upgrades and improved quality.

Contribution

It presents a task-agnostic reasoning architecture that separates planning from generation, with a novel two-stage training process for grounded, verifiable image synthesis.

Findings

Significant improvements in reasoning and image quality on text-to-image tasks.

Modular reasoning core allows for flexible upgrades without retraining entire models.

Two-stage training enhances plan grounding and visual correctness.

Abstract

Despite impressive progress in high-fidelity image synthesis, generative models still struggle with logic-intensive instruction following, exposing a persistent reasoning--execution gap. Meanwhile, closed-source systems (e.g., Nano Banana) have demonstrated strong reasoning-driven image generation, highlighting a substantial gap to current open-source models. We argue that closing this gap requires not merely better visual generators, but executable reasoning: decomposing high-level intents into grounded, verifiable plans that directly steer the generative process. To this end, we propose Unified Thinker, a task-agnostic reasoning architecture for general image generation, designed as a unified planning core that can plug into diverse generators and workflows. Unified Thinker decouples a dedicated Thinker from the image Generator, enabling modular upgrades of reasoning without retraining the entire generative model. We further introduce a two-stage training paradigm: we first build a structured planning interface for the Thinker, then apply reinforcement learning to ground its policy in pixel-level feedback, encouraging plans that optimize visual correctness over textual plausibility. Extensive experiments on text-to-image generation and image editing show that Unified Thinker substantially improves image reasoning and generation quality.