Bria Tailored Model Info

TailoredModelInfoNode (shown as Bria Tailored Model Info) doesn't generate anything. It's a lookup node - you give it a trained model's ID, it fetches that model's default settings from Bria's API and hands them back as outputs you wire into other nodes. Think of it as the "load checkpoint metadata" step for Bria's tailored models: the thing that stops you from guessing the right step count or prompt prefix by hand.

Bria's "tailored" models are the commercial-workflow analog of a LoRA: you train a small model on your own visual IP - a brand's product, a mascot, a specific art style - and then every generation with that model reproduces the look. The catch is that each trained model carries its own recommended generation settings, and they're not the same for every model. That's what this node is for.

How it works

You give it model_id (the ID Bria assigned when the tailored model finished training) and api_key, and it queries /v1/tailored-gen/models/<id>. It returns four outputs:

• generation_prefix - a text prefix that gets prepended to your prompt (this is how the model knows to invoke the trained style; feed it into ReimagineNode's tailored_generation_prefix or the Tailored Gen node's prompt).
• model_id - echoed straight back, handy for wiring downstream so you never have the ID typed in two places.
• default_fast (INT) and default_steps_num (INT) - the recommended speed/step settings for this specific model. The node derives them from the model's training_version: a light model gets fast=1 and 8 steps, a full model gets fast=0 and 30 steps.

Wire those two into the matching inputs on the generation node and you've replicated the settings the model was trained to run at, without memorizing them.

Installing it

Part of the BRIA AI API nodes pack (Bria-AI/ComfyUI-BRIA-API). ComfyUI Manager → search BRIA API → install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/Bria-AI/ComfyUI-BRIA-API

Restart ComfyUI; it's under API Nodes. No downloads, no GPU - you need a BRIA API token from platform.bria.ai. Note this node is a metadata query, not a generation, so it's effectively free - the billing happens when the output settings feed a real generation node.

Practical notes

It's a small node but easy to misuse. The model_id must be the exact ID of a trained model on your Bria account - a typo or a model owned by a different org gets you a 404-style error. And its real value is only unlocked if you actually route the outputs: a common pattern is TailoredModelInfoNode → ReimagineNode (prefix, fast, steps) for a styled reimagining, or → Bria Tailored Gen for a direct generation. If you've hardcoded the settings instead, you've missed the point - this node exists so the defaults can't drift when Bria updates a model.