Google Nano Banana Pro Text To Image (Channel Edition)

Nano Banana Pro is the bigger, heavier sibling of Google's Gemini image model - more fidelity, more patience with complicated prompts, still closed weights, still no local checkpoint to download. This node is the Channel Edition text-to-image wrapper, and it's the one I'd hand to a beginner: three real inputs, cloud execution, and the least room to screw up. If you've ever watched a friend's Nano Banana workflow die in a sampler settings rabbit hole, you get the appeal of "type prompt, pick resolution, queue."

Channel edition, explained

BizyAir ships every model twice: _Base_ classes hit BizyAir's own serving channel, _Official_ classes hit the vendor's official API surface. Same key, same gateway, different endpoint path (nano-banana-pro-base/text-to-image here). The Channel nodes strip the knobs down to what 90% of jobs actually need, which keeps pricing simple and mistakes rare. The Official nodes hand you seed, temperature, top_p and the rest - grab those when you need reproducibility or fine control, not when you're just trying to get a good image.

How it works

Queuing sends your prompt to api.bizyair.ai, the job runs remotely, and the result is downloaded back into ComfyUI as a native IMAGE tensor. No GPU load, no model files, no models/ folder. The flip side: your prompt leaves the machine and every job eats BizyAir credits, so it's worth getting your prompt right the first time.

Inputs that matter

• prompt - the whole job. Pro rewards descriptive natural language; it handles long, multi-clause prompts better than most.
• resolution - dropdown, defaults to 2K. Iterate lower, finish higher.
• aspect_ratio - defaults to 5:4 on this node, interestingly; change it to match your output canvas.
• api_config - per-request override of base URL and API key; only useful for testing alternate endpoints.
• skip_error - flip on and failures return a placeholder instead of killing the workflow.

Outputs are image, request_id, and urls. Wire image to a preview or save node; request_id is your receipt for support, urls gives you direct download links.

Install and get going

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

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

Restart ComfyUI, then make an account at bizyair.ai, grab the API key, and set it in Settings > BizyAirPlus > API Key (or as the BIZYAIR_API_KEY env var). Click the BizyAirPlus button in the action bar until it shows ON - nothing runs until that toggle is on.

Gotchas

The most common failure is also the most boring: missing key, toggle off, or the pack installed into the wrong Python. Verify with python -m pip show bizyair-cloudberry, check the ComfyUI console for the real error, and confirm the network can actually reach api.bizyair.ai. And remember you're paying per job on a hosted service - a little prompt polish up front beats four paid attempts.