Nodes/BizyAirPlus/Flux Kontext Max Text To Image (Channel Edition)
ComfyUI Node

Flux Kontext Max Text To Image (Channel Edition)

Flux Kontext Max text-to-image — BFL's 12B editing family, hosted, zero install

By siliconflow·Created 5 months ago·Updated about a month ago· 20
Flux Kontext Max Text To Image (Channel Edition)
  • api_config
  • image
  • request_id
  • urls
prompt
aspect_ratio21:9
skip_errorfalse

Flux Kontext is Black Forest Labs' 12B instruction-based generation family - the one that made "edit this image with a sentence" a local commodity and became famous for watermark removal and character-consistent edits. This node is the text-to-image channel of Kontext Max, which is the more streamlined tier: no image input at all, just a prompt and an aspect ratio, and the cloud returns a Kontext-generation image.

The name might confuse you if you only know Kontext as an editor, so let's be explicit: this T2I variant has no image input. It's pure prompt-to-picture. That's the whole "Text To Image" in the display name. The sibling I2I node is where the editing magic happens; this one is the generation side.

Inputs and outputs

  • prompt - required. Kontext is a T5-era model, so write natural language, not a comma soup. Full sentences describing the scene work noticeably better than tag-stack prompting.
  • aspect_ratio - dropdown, default 21:9. The ultra-wide default is Kontext Max's own preference - you can switch it, but the hosted default is telling you what the model is comfortable with.

Outputs: image (the generated result), request_id, and urls (JSON string of output file URLs). Standard BizyAir triple - wire the image into Save Image and you're done.

How it works

Like every BizyTRD_*_API node, this one is generated from BizyAir's model registry and calls the hosted flux-kontext-max-base/text-to-image endpoint. Your prompt and aspect ratio go up, the pack polls while the image renders, and the result downloads back into your graph. The key comes from BizyAirPlus settings or BIZYAIR_API_KEY unless you override it per-request. Because it's Kontext - the guidance-distilled Pro-descendant - you get BFL's quality without any local 12B download, which is the entire pitch of this pack.

Why you'd pick this over local Flux

Fair question, since you can run Flux locally. The honest answer: Kontext Max isn't the open-weights Dev tier - it's the hosted tier, and this node is the only way to get it without BFL's API. If you want Kontext-Max-class quality on a machine that can't run 12B, or you want the no-download one-node path, this is it. If you're already happily running Flux Dev locally for pure generation, this is a lateral move that costs credits - skip it and save the channel for the I2I edit nodes where hosted Kontext earns its keep.

Troubleshooting

  • Prompt ignored or washed out: check you're writing sentences, not tags. Kontext parses natural language; tag soup underuses it.
  • Red placeholder output: that's the pack's skip_error failure mode - the image output becomes a red canvas with the error text burned in. Check the console and the urls output for the real error and any original download links.

Install

Part of the BizyAirPlus pack:

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, make sure BizyAirPlus is ON and keyed, and try the 21:9 default before you fight it. Kontext Max picked that ratio for a reason.

CategoryBizyAirModelServices/BizyTRD/FLUX

Inputs (4)

NameTypeDefaultDescription
promptSTRINGPrompts
aspect_ratiooptCOMBO21:9Proportion
api_configoptBIZYAIR_OPENAPI_CONFIG单次请求覆盖base_url和api_key
skip_erroroptBOOLEANfalse开启后遇到错误不中断工作流,输出对应类型的错误占位符

Outputs (3)

NameTypeDescription
imageIMAGE
request_idSTRING
urlsSTRING