ComfyUI Extension Unifies Midjourney, GPT Image, Gemini in One Workflow

A new open-source ComfyUI custom node extension called ComfyUI-MidjourneyHub now lets AI artists and developers access 3 major commercial image generation models — Midjourney, OpenAI GPT Image, and Google Gemini — from a single, unified workflow interface. The project eliminates the long-standing pain point of juggling multiple API integrations and separate node packages to use different commercial AI drawing services.

For anyone who has spent time stitching together fragmented toolchains just to compare outputs from different models, this extension represents a significant quality-of-life improvement. One configuration, one node category, and access to the most powerful commercial image generation engines available today.

Key Takeaways at a Glance

• Unified access to Midjourney, OpenAI GPT Image, and Google Gemini image generation within ComfyUI
• Supports text-to-image, image editing, upscaling, variations, blending, and batch operations
• Already compatible with OpenAI's latest gpt-image-2-all model
• Gemini integration uses Gemini 3 Pro Image Preview for generation and editing
• All nodes organized under a single 'MidjourneyHub' category for easy drag-and-drop workflow building
• API calls routed through a cloud proxy service — configure once, use all models

The Fragmentation Problem This Extension Solves

ComfyUI has become the go-to node-based interface for AI image generation workflows, particularly among power users who want granular control over their pipelines. Its modular architecture allows custom nodes to extend functionality almost infinitely. But that flexibility comes with a cost: fragmentation.

Until now, using commercial models like Midjourney or GPT Image inside ComfyUI meant hunting down separate custom node packages for each service. Each package had its own API configuration requirements, its own node naming conventions, and its own quirks. Switching between models in a single workflow was tedious at best and broken at worst.

ComfyUI-MidjourneyHub tackles this head-on by providing a single extension that wraps all 3 commercial services under one consistent interface. The developer, who shared the project with the open-source community, described the motivation simply: 'Everyone doing AI art has this pain point — wanting to use Midjourney, GPT Image, and Gemini in ComfyUI workflows, but each requires its own node and API setup. The experience is fragmented.'

Full Node Breakdown: What You Can Do

The extension ships with a comprehensive set of nodes covering the core capabilities of each supported model. Here is what is currently available:

Midjourney Nodes

• Imagine — Standard text-to-image generation using Midjourney's engine
• Upscale — High-resolution upscaling of generated images
• Variation — Create variants of existing outputs
• Batch — Run multiple generation jobs simultaneously
• Blend — Merge and fuse multiple images together

OpenAI GPT Image Nodes

• Generate — Text-to-image creation powered by OpenAI's image models
• Edit — Modify and transform existing images with text prompts
• Full support for the latest gpt-image-2-all model, which OpenAI released as its most capable image generation model to date

Google Gemini Nodes

• Generate — Text-to-image generation using Gemini's multimodal capabilities
• Edit — Image editing and manipulation via Gemini
• Built on Gemini 3 Pro Image Preview, Google's latest image-capable model variant

All nodes sit under a single 'MidjourneyHub' category in the ComfyUI node browser, making them easy to find and drag into any workflow without searching through dozens of unrelated extensions.

Why This Matters for AI Image Workflows

The real value of this extension goes beyond mere convenience. It fundamentally changes how creators can approach model comparison and multi-model pipelines inside ComfyUI.

Consider a common scenario: an artist wants to generate a concept using Midjourney's distinctive aesthetic style, then refine specific elements using GPT Image's editing capabilities, and finally run a variation through Gemini to explore a different interpretation. Previously, this required 3 separate extensions, 3 API configurations, and careful manual handoff between nodes that were never designed to work together.

With ComfyUI-MidjourneyHub, this becomes a single coherent workflow. The unified node design means outputs from one model's generation node can flow directly into another model's editing node. This kind of cross-model pipeline is increasingly important as each commercial service develops distinct strengths.

Midjourney remains the gold standard for artistic, stylized imagery. GPT Image excels at instruction-following and precise edits. Gemini brings Google's multimodal reasoning to the table, offering unique capabilities in understanding complex scene descriptions. Having all 3 accessible in one workflow unlocks creative combinations that were previously impractical.

Technical Architecture: Proxy-Based API Access

One notable architectural decision is the use of a cloud proxy API to route all model calls. Rather than requiring users to configure direct API credentials for each service individually — which in Midjourney's case is particularly complex since it lacks a straightforward public API — the extension channels requests through an intermediary proxy service.

This approach has several practical advantages:

• Single API key configuration — set it up once and access all 3 services
• Simplified authentication — no need to manage separate OAuth flows or API tokens for each provider
• Consistent error handling — the proxy layer normalizes responses across different backend services
• Midjourney access without Discord — traditionally, Midjourney API access requires workarounds through Discord bot interactions or third-party reverse-engineered endpoints

However, users should be aware that proxy-based access means routing requests through a third-party service. This introduces considerations around data privacy, latency, and pricing that differ from direct API access. The proxy service handles billing and rate limiting, which means costs are determined by the proxy provider rather than the underlying model providers directly.

For professional workflows where data sensitivity is a concern, this architecture warrants careful evaluation. For hobbyists and independent creators, the convenience likely outweighs the tradeoffs.

How This Fits Into the Broader AI Image Generation Landscape

The release of ComfyUI-MidjourneyHub reflects a broader trend in the AI creative tools ecosystem: consolidation and interoperability. As the number of capable image generation models has exploded — from Stable Diffusion and DALL-E to Midjourney, Flux, Imagen, and beyond — creators increasingly need tools that bridge these services rather than lock them into one.

ComfyUI itself has been at the center of this trend. Unlike simpler interfaces like Automatic1111's WebUI, ComfyUI's node-based architecture is inherently designed for complex, multi-step workflows. The community has built thousands of custom nodes, but the ecosystem has grown organically and often chaotically.

Projects like MidjourneyHub represent a maturation of this ecosystem. Instead of building narrow, single-purpose nodes, developers are creating integration layers that treat multiple AI services as interchangeable components in a larger pipeline. This mirrors what is happening in the LLM space with tools like LiteLLM and OpenRouter, which provide unified APIs across dozens of language model providers.

The AI image generation market itself continues to heat up. OpenAI's GPT Image models have gained significant traction since their release, while Google's Gemini image capabilities — particularly with the Gemini 2.0 and 3.0 series — are rapidly closing the gap with competitors. Midjourney, despite its unconventional Discord-first distribution model, remains a favorite among digital artists for its distinctive visual quality.

What This Means for Developers and Creators

For ComfyUI power users, the extension is an immediate productivity boost. No more maintaining multiple node packages or debugging incompatible API configurations. The unified interface reduces setup friction and makes it practical to experiment across models without context-switching overhead.

For developers building AI image applications, the project offers a reference architecture for how to design multi-provider integrations. The proxy-based approach, while not suitable for every use case, demonstrates a clean pattern for abstracting away provider-specific complexity.

For studios and creative teams, the ability to run comparative workflows — generating the same prompt across Midjourney, GPT Image, and Gemini simultaneously — could streamline creative direction decisions and reduce time spent manually testing different services.

Looking Ahead: The Future of Multi-Model Creative Workflows

The trajectory here is clear. As commercial image generation models continue to improve and differentiate, the demand for unified workflow tools will only grow. We can expect to see more ComfyUI extensions that aggregate commercial services, and potentially more sophisticated features like automatic model selection based on prompt characteristics or cost optimization across providers.

The developer has indicated the project is actively maintained, with support for new models being added as they become available — the quick adoption of gpt-image-2-all is evidence of this responsiveness. Future additions could potentially include services like Flux Pro, Ideogram, or Adobe Firefly as those platforms expand their API offerings.

For now, ComfyUI-MidjourneyHub fills a genuine gap in the creative AI toolchain. It is available as an open-source project and can be installed through standard ComfyUI custom node installation methods. Anyone running ComfyUI who has been frustrated by the fragmented state of commercial model integration should find it worth a look.