VolcengineChatOption(NYJY)
Pick a Doubao chat model by ticking a box
- model
The model field on VolcengineChat is a plain text input, which is flexible and also a footgun - one typo and you're calling a model ID that doesn't exist. VolcengineChatOption is the pack's answer: instead of typing a Doubao model ID, you tick a checkbox and the node hands you the right string. It's a dropdown in disguise.
It's a helper node from ComfyUI_NYJY, the same aidenli pack that wraps the Volcengine (ByteDance Volcano Engine) API for chat and image generation. If you've got a workflow where you flip between chat models to compare outputs, this keeps the model choice visible and one click away instead of buried in a text field.
How it works. Three boolean inputs, one per model, all defaulting to off:
doubao-seed-1-6-251015- the main Seed 1.6 chat modeldoubao-seed-1-6-vision-250815- the vision-capable one that pairs withVolcengineChat's image inputdoubao-seed-1-6-flash-250828doubao-seed-1-6-thinking-250715- and here's where you need to look twice.
The trap. That last label isn't one model - it's two model IDs jammed together with no separator: the flash model and the thinking model share a single checkbox. Tick it and the node outputs the whole mashed string, which is not a valid Ark model ID, and VolcengineChat will fail or return garbage. So treat that checkbox as "don't tick this one" unless you're willing to edit the output after the fact. It's a genuine rough edge in the pack, the kind of thing you only notice when the model call quietly dies.
The output. A single model string: the label of whichever checkbox is on. Wire it straight into VolcengineChat's model input. A couple of behaviors to know: if more than one box is on, the first ticked one in the node's field order wins, which is easy to lose track of; if none are on, you get an empty string, and an empty model ID is exactly the kind of thing that produces an unhelpful API error. Keep exactly one ticked.
Why bother? Because swapping models becomes a UI action instead of a text edit. Want to compare Seed 1.6 vs. the vision build on the same prompt? Toggle, run, toggle back. It also documents itself - the workflow shows at a glance which model it's wired to. For the price of one checkbox node, that's a reasonable trade for anyone doing model-A/B runs.
Install. It comes with the whole pack:
cd ComfyUI/custom_nodes
git clone https://github.com/aidenli/ComfyUI_NYJY
pip install -r requirements.txt
(ComfyUI Manager → "Install via Git URL" → https://github.com/aidenli/ComfyUI_NYJY also works.) The node itself is pure Python, but you'll need the Volcengine SDK installed for the chat node it feeds, and the same Volcano Engine API key applies - see the VolcengineChat article for the sign-up reality check.
When to reach for it. When you switch Doubao models often and you're tired of retyping - or when you want a workflow that makes its model choice obvious to anyone who opens it. Just avoid that double-name checkbox.
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| doubao-seed-1-6-251015 | BOOLEAN | false | — |
| doubao-seed-1-6-vision-250815 | BOOLEAN | false | — |
| doubao-seed-1-6-flash-250828doubao-seed-1-6-thinking-250715 | BOOLEAN | false | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| model | STRING | — |