FLUX 1.1 [pro]

FLUX 1.1 [pro] is the model between the two generations everyone talks about. FLUX 1.0 was the open-weights takeover; FLUX 2 was the 2025 flagship that ruffled feathers over size and censorship. FLUX 1.1 is the quiet one in the middle - API-only, faster than 1.0 pro, and it never got the community fuss that either neighbor did. If you've seen "FLUX 1.1 [pro]" on an image and wondered what it was, this is the node that answers.

The most visible sign you're dealing with the newer generation: there is no steps input. Look at the node and it's gone. Where the 1.0 pro node hands you a steps slider running to 100, the 1.1 endpoint has the sampling budget baked in server-side. BFL tuned it to produce its results at a fixed, faster cost - you get quality without the ability to throw more compute at it. That's the whole 1.1 design philosophy: fewer dials, better defaults, faster turnaround.

Inputs that matter

• prompt (required) - natural language; as always with FLUX, sentences beat tag soup.
• guidance (default 2.5, range 1.5–5) - the one knob that still does real work here.
• width / height (up to 1440, step 32) - explicit resolution control.
• interval (1–10, default 1) - guidance interval; leave it unless you know better.
• prompt_upsampling (default true) - the API rewrites your prompt before generating. Flip it off if you want your wording kept verbatim.
• image_prompt (optional IMAGE) - remix an existing image into the generation.
• safety_tolerance (0–6, default 2).

Output is a single IMAGE tensor - Save Image, Preview, or wire it onward. Nothing exotic about the shape of it.

How it fits the pack

Within bfl-comfy-nodes, the 1.1 pair is the "modern default" tier: FLUX 1.1 [pro] when you want control, FLUX 1.1 [ultra] when you want zero dials. Compared to 1.0 pro, this node is cheaper per image at the same settings because the step count is fixed rather than cranked. Compared to ultra, it still lets you set resolution and guidance - so it's the one people actually build workflows around, while ultra is the drop-in-and-forget.

Install

Same one-time pack setup: ComfyUI Manager → Black Forest Labs API Nodes, or git clone https://github.com/black-forest-labs/bfl-comfy-nodes into custom_nodes, restart, and set BFL_API_KEY (env var, bfl_api_key.txt beside the pack, or the api_key_override field). Key from api.bfl.ml. No model downloads, no heavy dependencies.

The honest caveats are the pack's shared ones: it's a paid call, moderation sits in front (safety_tolerance is your only lever there), and there's a 240-second poll timeout if BFL's servers are slow. But for a "give me good, fast, and I don't want to count steps" default, this is the node most people in this pack actually end up using day to day.