FLUX.2 [max]
The 32B flagship that doesn't need your VRAM
- referenceImages
- advancedFeatures.watermark.image
- image
FLUX.2 [max] is the most capable image model BFL has ever shipped - a 32-billion-parameter rectified flow transformer with up to 4-megapixel output and a dozen-or-so quality points over FLUX.1. The catch, if you've priced a GPU lately: it needs something like 18–24GB of VRAM even quantized, and running it locally is a whole machine's worth of commitment. The Runware route through this node just… skips that. The bfl:7@1 AIR is baked in; the denoising happens on their fleet.
There's a real irony the KB likes to point out: FLUX.2 won the benchmarks and still lost the community to Alibaba's smaller, Apache 2.0, uncensored Z-Image. On the API, though, you get the benchmark winner without the argument - this is the pure "I want the best FLUX.2 quality, now" node.
How it works
A taskType: imageInference REST call. The genuinely cool part: FLUX.2 generation and editing live in one model, and reference images ride in via the referenceImages socket (up to 10 of them). You can prompt it into both generating from scratch and editing from references with the same node.
Inputs that matter
- positivePrompt - required; everything else is optional.
- referenceImages - the differentiator. Feed one or several images and describe what to do with them. This is what turns a generation node into an editing node.
- width / height - 256–2048 in 16px steps. Native 4MP territory is here; go big if your use case wants it.
- providerSettings.bfl.promptUpsampling - prompt expander.
- providerSettings.bfl.safetyTolerance - 0–5, default 2; the BFL moderation dial.
- seed / numberResults - reproducibility and variants.
Install
cd ComfyUI/custom_nodes
git clone https://github.com/Runware/ComfyUI-Runware
pip install -r ComfyUI-Runware/requirements.txt
Or ComfyUI Manager → search Runware → Install → restart. API key in ComfyUI Settings → Runware API key or RUNWARE_API_KEY. Prepaid top-up; per-run cost on the title bar.
Gotchas
- The max tier is the priciest image node in the pack. Quality is real, but it's the wrong tool for bulk iteration - draft cheap, finalize here.
- Moderation is BFL-side;
safetyToleranceis your only knob. - No steps/CFG widgets - provider-fixed, turnkey. If you want the dials, the pack's klein node has them.
FLUX.2 [max] is the "no compromises, no hardware" answer in this pack. It's also the answer the community quietly stopped choosing locally - but on the API, with nothing to download and nothing to license-wrangle, it's the cleanest way to touch the actual flagship.
Inputs (22)
| Name | Type | Default | Description |
|---|---|---|---|
| positivePrompt | STRING | Text prompt describing elements to include in the generated output. | |
| referenceImagesopt | IMAGE | — | |
| advancedFeatures.watermark.imageopt | IMAGE | — | |
| widthopt | INT | 1024256–2048 | Width of the generated media in pixels. |
| heightopt | INT | 1024256–2048 | Height of the generated media in pixels. |
| seedopt | INT | 00–9223372036854776000 | Random seed for reproducible generation. When not provided, a random seed is generated in the unsigned 32-bit range. |
| numberResultsopt | INT | 11–20 | Number of results to generate. Each result uses a different seed, producing variations of the same parameters. |
| advancedFeaturesopt | BOOLEAN | false | Enable to set advancedFeatures. Off uses the model's default. |
| advancedFeatures.watermark.bgColoropt | STRING | Background color in hex format. | |
| advancedFeatures.watermark.displayPositionopt | COMBO | (default) | Watermark position. |
| advancedFeatures.watermark.fontColoropt | STRING | Text color in hex format. | |
| advancedFeatures.watermark.opacityopt | BOOLEAN | false | Enable to set advancedFeatures.watermark.opacity. Off uses the model's default. |
| advancedFeatures.watermark.opacity_valueopt | FLOAT | 0.100.1–1 | Watermark opacity. |
| advancedFeatures.watermark.textopt | STRING | Watermark text. | |
| providerSettings.bfl.promptUpsamplingopt | BOOLEAN | false | Enable automatic enhancement and expansion of the input prompt. |
| safetyopt | BOOLEAN | false | Enable to set safety. Off uses the model's default. |
| safety.checkContentopt | BOOLEAN | false | Enable or disable content safety checking. |
| providerSettings.bfl.safetyToleranceopt | INT | 20–5 | Tolerance level for content moderation. Lower values are stricter; higher values are more permissive. |
| ttlopt | BOOLEAN | false | Enable to set ttl. Off uses the model's default. |
| ttl_valueopt | INT | 60 | Time-to-live (TTL) in seconds for generated content. Only applies when `outputType` is `URL`. |
| outputFormatopt | COMBO | JPG | File format for the generated image. |
| outputQualityopt | INT | 9520–99 | Compression quality of the output. Higher values preserve quality but increase file size. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | — |