☁️BizyAir GPT Image 1 Text To Image

GPT Image 1 is OpenAI's current image-generation model - API-only, shipped April 2025, the one that superseded DALL-E 3 with noticeably better text rendering and native multimodal input support. This node gets it into a ComfyUI graph - composited with everything else you're building - without you having to separately set up an OpenAI API key and billing account. You pay through your existing BizyAir credits instead.

How it works

Two things worth knowing before you plug a prompt in. First, size here is an aspect ratio, not a pixel dimension - the backend picks the actual resolution for whichever ratio you choose. Second, this is a straight relay: OpenAI's own content moderation applies to whatever you send, the same filters that govern ChatGPT's image tool. Community reports through 2025 describe that policy moving in both directions over time - guardrails loosened in March 2025 (the update that triggered the viral Ghibli-style portrait wave), then tightened back up at points afterward - so don't assume a prompt that worked last month will still work today. That's upstream of BizyAir; there's nothing on the relay side to adjust for it.

The inputs and outputs that matter

• prompt - required, multiline.
• size (default 1:1; choices 1:1, 2:3, 3:2) - aspect ratio, not a resolution.
• variants (default 1; choices 1, 2, 4) - how many images to generate per call. Fixed enum, not a free integer.

Outputs: images, plus bizyair_model_name (mapping to gpt-image-1).

How to install it

ComfyUI Manager: search BizyAir, install, restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAir

then restart. This is a paid cloud call - set your API key on first use per the README and expect it to draw from your BizyAir credits per generation.

Common issues & troubleshooting

Prompt gets refused or silently altered. That's OpenAI's own moderation applying server-side, not a BizyAir bug - and per community reports the strictness of that filtering has genuinely shifted over time, so a prompt that used to work isn't guaranteed to keep working.

Wanting more than 4 images per call. Not available - variants is capped at the 1/2/4 enum. Queue multiple calls if you need more.

Need to edit an existing image instead of generating fresh. Wrong node - this one is text-to-image only, with no image input at all. Use the sibling BizyAir GPT Image Edit node for edits against an existing image.