Qwen Image Text To Image

Before the 2.0 API-only pivot, Qwen-Image was the 20B Apache 2.0 model that made the community trust Alibaba's image line - the one whose Edit variant became the open editing standard and whose open weights you can still run locally. This node is the original qwen-image/text-to-image endpoint, and unlike the 2.0 nodes it ships with diffusion-style controls: width, height, guidance_scale, and steps. It's the closest thing in the pack to the sampler-tuning feel of a local workflow, just executed on BizyAir's cloud. Given the model is Apache 2.0 and locally runnable, this node's main draw is convenience - no GPU, no download - or using it as a reference point against your local Qwen-Image setup.

How it works

Same cloud machinery as every BizyTRD node: your prompt posts to api.bizyair.ai, the job runs remotely, and the image is downloaded back into ComfyUI as a native IMAGE tensor. The difference is the parameter philosophy. The 2.0 nodes hide the diffusion internals behind a size dropdown; this one hands you the actual knobs, because the original Qwen-Image API exposed them.

Inputs that matter

• prompt - the positive prompt. Original Qwen-Image is happiest with descriptive natural language.
• negative_prompt - what to avoid; genuinely useful here since you have real guidance control to pair it with.
• width / height - integers, both default 1024, range 256–2048. Pick your canvas.
• guidance_scale - default 4.0, range 0.1–10. Higher = prompt adherence, lower = more creative drift. This is the single knob worth actually tuning.
• steps - default 15, range 6–50. More steps cost more credits; the model converges fine at the default for most subjects.
• seed - integer, default 0, for reproducible runs.
• api_config / skip_error - the pack standard: per-request key/base override and placeholder-on-error.

Outputs: image, request_id, urls.

Install and get going

Install through ComfyUI Manager (search "BizyAirPlus") or:

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

Restart, register at bizyair.ai, put the key in Settings > BizyAirPlus > API Key (or BIZYAIR_API_KEY), and click the BizyAirPlus action-bar toggle to ON. No toggle, no requests.

Gotchas and honest takes

Setup failures are the usual trio: missing key, toggle off, wrong Python (python -m pip show bizyair-cloudberry). One honest caveat about why you'd use this node: since the original Qwen-Image weights are open and Apache 2.0, a local run gives you the same model with no per-job billing - this node is for when you don't have the hardware or just want a zero-download baseline. And because you're paying per job, resist the urge to brute-force with high steps; find the quality you need at the default and save your credits.