BailianChatOption(NYJY)

This is a one-job utility node: it's a model picker. Where BailianChat makes you type a model ID into a text field, BailianChatOption gives you a checkbox per model and outputs the name of whichever one you ticked, ready to wire into BailianChat's model input. Nothing more, nothing less. If you only ever use one model you can skip it entirely and just type qwen-plus - this node exists to make a workflow readable, not to add capability.

How it works

The node is just a list of ten booleans, one per model the author decided to expose:

• qwq-plus and qwq-32b - the QwQ reasoning models (long-thinking, show-their-work style replies)
• qwen-max - Alibaba's biggest general chat model
• qwen-plus - the default workhorse
• qwen-long - built for long-context inputs
• deepseek-r1, deepseek-v3, deepseek-r1-distill-llama-70b, deepseek-r1-distill-qwen-32b - DeepSeek models served through the Bailian platform
• qwen-omni-turbo - the multimodal one (text in, but also audio/video understanding on the platform side)

Tick one, and it outputs the model's ID as a STRING. Tick more than one and the node still just picks based on the order of the list - in practice you'll want exactly one checked. All of these are served through the same Bailian (DashScope) endpoint that BailianChat calls, so the key, billing, and real-name-verification requirements from that article apply unchanged.

The one output

model - a STRING holding the model ID, e.g. qwen-plus. Feed it directly into BailianChat's model field. That's the entire contract. There are no optional inputs and no other outputs.

Installing

Part of the NYJY pack, so: ComfyUI Manager → Install via Git URL → https://github.com/aidenli/ComfyUI_NYJY, restart. No model downloads, no extra config for this node itself - the API key lives in config.json (or on the BailianChat node).

When it's actually worth it

Honest take: for a single fixed workflow, type the model name and move on. Where BailianChatOption earns its keep is when you're building a template you hand to other people - a shared workflow where you want the person downstream to see all the available models and pick with a click instead of guessing an ID. It's the difference between a dropdown and a magic string, and for sharing, a few checkboxes beat "what's the exact model ID again?" every time. It's also handy if you're A/B-ing models: swap the checked box and re-run to compare qwen-plus against qwen-max output without touching any text field.