Nodes/comfyui-budgetpixel/Grok Imagine (BudgetPixel)
ComfyUI Node

Grok Imagine (BudgetPixel)

XAI's model with the aesthetics reputation, at a flat rate

By budgetpixel-ai·Created about 17 hours ago·Updated about 12 hours ago· 3
Grok Imagine (BudgetPixel)
  • reference_images
  • images
prompt
aspect_ratio1:1
num_images1
size1K
force_rerunfalse

Grok's image model built its reputation inside X/Twitter, where people kept noticing that its output looked suspiciously good - the community even clocked early on that it appeared to lean on FLUX-family weights. Whatever the lineage, the thing people actually report is aesthetics: Grok Imagine has a look, and it's a look people like. This node is xAI's model, hosted, on the BudgetPixel shelf, at a flat 35 credits per image whether you render at 1K or 2K.

The flat pricing is the quietly good part. Most of the pack's image nodes make you think about resolution tiers and quality multipliers; Grok Imagine doesn't. 35 credits, one image, either resolution - and it does instruction-based editing from up to three reference images on top of text-to-image. It's the "I want a good-looking image without reading a billing matrix" node.

How it works

Standard BudgetPixel machinery: POST /images/grok-imagine-image, poll /images/{id}, download the IMAGE batch. The editing path is instruction-driven - you wire in up to three reference IMAGE connections and write the change as natural language in the prompt ("turn the background into a rainy city street, keep the lighting moody") rather than expecting masks or parameters. If you've used a closed image API before, this is that pattern: the model does the reasoning, you do the describing.

The inputs that matter

  • prompt - required, multiline. With references present, this is the edit instruction.
  • reference_images - up to 3 for editing and composition.
  • size - 1K or 2K. No quality multiplier, no credit difference - pick 2K when you'll crop or print, 1K when it's for the screen.
  • aspect_ratio - for framing.
  • num_images - 1–4 (each multiplies the 35-credit bill).
  • seed - default -1 (random).

Output is images (IMAGE batch).

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. requests is the only dependency; no model files.

Where people get burned

The aesthetics that made Grok famous are exactly why moderation expectations matter here - xAI's filter is famously the lightest of the closed labs (the modidex closed-source-models doc puts it at "least restrictive," with the caveat that it still moderates). Don't assume "light" means "off." Second, this node is strictly image - the Grok image model, not the chatbot, not video; people searching "Grok" hoping for the whole package should look at the video shelf too. Third, flat pricing is a trap of its own: because 2K costs the same as 1K, it's tempting to always hit 2K, which is slower server-side and heavier downstream. For comping and drafts, 1K is fine and your render pipeline will thank you.

CategoryBudgetPixel/Image

Inputs (6)

NameTypeDefaultDescription
promptSTRINGText description of the image to generate.
aspect_ratiooptCOMBO1:1Output aspect ratio.
num_imagesoptINT11–4Number of images to generate.
reference_imagesoptIMAGEOptional reference images (up to 3) for image editing. Each item is a public image URL, a data URI, raw base64, or an uploaded-file URL from POST /v1/uploads. Omit for text-to-image. Use aspect_ratio "match_input_image" (or "auto") to let the model follow the input.
sizeoptCOMBO1KOutput resolution tier. Both tiers bill the same 35 credits per image.
force_rerunoptBOOLEANfalse

Outputs (1)

NameTypeDescription
imagesIMAGE