OpenAI GPT Image 2 Text To Image (Channel Edition)

This is the text-to-image member of the GPT Image 2 family in BizyAirPlus, in its Channel Edition - which means the simplest version of a simple thing. One required input, prompt. Type it, queue it, get an image back. If you want OpenAI's image model inside ComfyUI without any of the local-model ceremony, this is the lightest possible way to get it.

GPT Image is OpenAI's successor line to DALL-E, and the thing the community actually respects it for is text rendering - clean, legible text inside generated images, which is still the benchmark open models chase. If your workflow needs signage, packaging, or any image with words in it, this is a legitimate reason to reach for a closed API node.

Why you'd reach for it

Because there's a whole class of jobs where your local Stable Diffusion or Flux stack is the wrong tool. Text-in-image is the big one, but there's also just the convenience of a billed, always-available model that doesn't depend on what you have installed. The Channel Edition deliberately keeps the controls minimal: resolution defaults to 1K and aspect_ratio defaults to auto. No width/height math, no pixel-count constraints - the model decides. If you want to micromanage dimensions, the Official Edition of this node is the one with the sliders.

How it works

Wire the prompt in, optionally pick resolution and aspect_ratio, queue. The node submits the request through BizyAir's channel to GPT Image 2, polls until the render completes, and returns three outputs: image (IMAGE - connect it to Save Image or a preview node like any local generation), request_id (STRING - trace ID), and urls (STRING - where the output file lives). There is no local model, no VRAM, no seed control, no negative prompt. What you type is the whole creative lever.

Two optional inputs keep it workflow-safe: api_config (BIZYAIR_OPENAPI_CONFIG) overrides base_url and api_key for one request, and skip_error makes a failed call output an error placeholder instead of aborting the queue.

Install and setup

Same one-time install as every node in this pack:

cd ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAirPlus.git
cd BizyAirPlus
python -m pip install -r requirements.txt

Restart ComfyUI (or install "BizyAirPlus" via ComfyUI Manager). No model downloads - the bizyair-cloudberry and bizytrd pip packages auto-install on startup. Then click the BizyAirPlus button in the action bar to ON, create an account at bizyair.ai, enter the API key when prompted (or in Settings > BizyAirPlus > API Key), and queue.

Common issues

• No image comes back - switch off or key missing are the usual culprits; the ComfyUI console has the detailed error.
• Text looks off - GPT Image is good at text but the prompt has to be explicit about what it says and where. Spell it out in quotes.
• Content filters - OpenAI's image models carry heavier guardrails than most, and the community has documented them swinging between strict and looser. If a request is refused, no input on this node changes that.

Pack-level caveat, stated once: BizyAir auto-installs and sends your prompts to the cloud. It's a legitimate SiliconFlow product, but flip the switch OFF for local-only work and treat every prompt as something being sent to a server.