FLUX 2 Pro (BudgetPixel)
The megapixel flagship when you need the BFL ceiling
- image
- images
FLUX 2 Pro is BFL's premium API tier, and in this pack it's the "stop messing around, I need the good one" button. Where the Dev build is about multi-reference editing, Pro is about raw output quality at big resolutions: it's the node that goes to 4 megapixels, and at 25 credits per generation it undercuts what you'd expect a flagship hosted render to cost. If you've ever wished the open-weights FLUX 2 you run locally could just... be a little bigger, this is the answer that doesn't touch your VRAM.
It's also the honest endpoint of a decision you should make deliberately: this node and the 25-credit FLUX 2 Dev are the same price, so the real question is which job you're doing. Multi-reference composition → Dev. Single-image, highest-possible-detail rendering → Pro. They're siblings, not upgrades of each other, and the pack's pricing makes that visible.
How it works
Same BudgetPixel machinery as every image node - POST /images/flux-2-pro, poll /images/{id}, download the IMAGE batch. The thing that makes Pro distinct is size, a megapixel enum (0.5MP, 1MP, 2MP, 4MP). That's the resolution ladder, and it's the difference between "a good image" and "a poster." The 4MP tier is where Pro earns its keep: a local FLUX 2 can do 4MP in principle, but you're renting that capability for 25 credits instead of occupying your GPU for the better part of a coffee break.
The inputs that matter
- prompt - required, multiline.
- size - the megapixel enum. This is the node's defining knob: start at 1MP for drafts, jump to 4MP for finals.
- image - optional. Feed an
IMAGEfor single-image editing / image-to-image; the prompt describes the change. - aspect_ratio - 7 options (1:1, 4:3, 3:4, 9:16, 16:9, 2:3, 3:2).
- num_images - 1–4, default 1 (multiplies credits per call).
- seed - default -1 (random) for reproducible finals.
Output is images (IMAGE batch). It feeds whatever your pipeline does next - Save Image, an upscaler, an img2img pass.
Install
Manager (search "BudgetPixel") or:
cd ComfyUI/custom_nodes
git clone https://github.com/budgetpixel-ai/comfyui-budgetpixel.git
pip install -r comfyui-budgetpixel/requirements.txt
Restart, set BUDGETPIXEL_API_KEY or budgetpixel.json ({"api_key": "bpx_live_..."}) in the pack folder. Only requests as a dependency, no model files.
Where people get burned
The mistake that actually costs money here is using the 4MP tier for everything. The node's own price is flat regardless of size, but a 4MP generation is slower server-side and heavier downstream - if you're feeding the output into an upscaler anyway, 2MP is usually the sweet spot and you've wasted nothing. Second: Pro is hosted, which means BFL's moderation is in the loop and no local workflow hack bypasses it - the exact trade discussed across the API-node ecosystem. Third, keep seed set once you like a result; a random seed each run is how "I liked that one" becomes unreproducible, and the 25 credits get spent finding the same image twice. And if you want to re-run deliberately, that's what force_rerun is for - the cache won't spend your credits on a no-op.
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | Text description of the image to generate. | |
| aspect_ratioopt | COMBO | 1:1 | Output aspect ratio. |
| imageopt | IMAGE | Optional image-to-image input. Provide a public image URL, a data URI, raw base64, or an uploaded-file URL from POST /v1/uploads. Omit for text-to-image. | |
| num_imagesopt | INT | 11–4 | Number of images to generate. |
| seedopt | INT | -1-1–2147483647 | Seed for reproducible generation. Omit for random. |
| sizeopt | COMBO | 1MP | Output resolution (megapixels). |
| force_rerunopt | BOOLEAN | false | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| images | IMAGE | — |