zhenzhen-wan-2.7-global-image-lowprice

Wan's numbered image models don't exist as weights you can download - 2.7 is API-only, like 2.5 and 2.6 before it. Comfly_wan_2_7_global_image_lowprice is the node that gets you onto the "global" tier of that API cheaply, from Zhenzhen's domestic storefront. The model dropdown is the whole node: wan-2.7-global-t2i for text-to-image, and wan-2.7-global-i2i / wan-2.7-global-i2i-pro for image editing with references. Three modes, one node, one price.

It's part of the low-price family that runs on api.seedance.nz, so like its siblings it needs the shared settings node rather than a direct key. Everything runs in the cloud; your GPU is on vacation.

How it works

Pick a model, write a prompt, choose your canvas, and the node submits the job and returns the finished image plus image_url, task_id, and response. For i2i modes, connect reference images to image1–image9 and the prompt describes what changes.

The inputs that do the work:

• model - the t2i / i2i / i2i-pro choice. Pro is the stronger editor and presumably pricier; try it when the base i2i under-delivers.
• prompt - the creative brief. With reference images, say what to change; otherwise, say what to make.
• width / height - 512 to 4096, stepped by 8. Unusually flexible for a cloud image node, so you can hit non-standard canvases instead of being locked to ratios.
• thinking_mode - on by default. The model reasons about the prompt before generating, which helps with complex or multi-part requests. It costs a little more; flip it off for simple prompts.
• image1–image9 - references for i2i.

Installing and setting up

Install through ComfyUI Manager (search Comfyui-zhenzhen) or:

cd ComfyUI/custom_nodes
git clone https://github.com/T8mars/Comfyui-zhenzhen

Restart ComfyUI. No model downloads. This node takes no direct key - connect the Comfly Seedance2 Low Price Settings node's api_config output, or set SEEDANCE_API_KEY. Same settings node feeds the whole low-price family.

Troubleshooting

The universal family stumble is the config: run it without an api_config connection and you get an "empty api_key" error. Set the key once in the settings node and reuse it. Beyond that, two Wan-image quirks: very wide or tall canvases (approaching 4096) can produce wonky compositions, and thinking_mode off with a long prompt is where quality falls off a cliff. Keep thinking on for anything with multiple subjects or text. And skip_error is available if you'd rather a failed image return empty than abort a batch.

Where people get surprised: the "global" in the name isn't a style - it's ByteDance's model tier for non-mainland users, so it's the one to reach for if you're outside China and want the latest Wan image without fighting a domestic endpoint.