Replicate black-forest-labs/flux-1.1-pro
The faster, better top Flux tier
- IMAGE
Flux 1.1 Pro is the upgraded top tier of the Flux family - faster and higher quality than the original Flux Pro, and the one that topped image-generation leaderboards when it dropped (the model everyone was calling "blueberry" before it was named). Like Pro, it's API-only: no downloadable weights, so this node is how you reach it from ComfyUI. It runs on Replicate's cloud, so you need a Replicate account and a REPLICATE_API_TOKEN, you pay per image, and there's nothing to install or run locally.
Straight recommendation: if you were going to use flux-pro, use this instead. It's the newer generation - generally faster, cheaper per image, and better - which is why its control surface is deliberately slimmed down. BFL tuned it so you don't have to.
How it works
1.1 Pro is tuned to produce its best output without you fiddling with step counts or guidance, so those knobs aren't exposed here at all. You get the prompt, size, and a couple of switches. As with every Flux tier there's no negative prompt - that's a Flux-wide trait, not a limitation of this node.
The inputs and outputs that matter
prompt(required) - natural language; Flux handles full sentences and complex scenes well.aspect_ratio(1:1) - a ratio like16:9,2:3,9:16, orcustomif you want exact dimensions.width/height(256–1440) - only used whenaspect_ratiois set tocustom; otherwise the ratio drives the size.prompt_upsampling(false) - lets the model expand your prompt with an LLM for richer detail. Handy for a lazy one-liner, but it takes control away from you, so leave it off when you want your prompt used literally.safety_tolerance(2, range 1–5) - how permissive the filter is; higher is more lenient but capped.
Output is a single IMAGE. output_format/output_quality set the file, seed fixes the result, and force_rerun bypasses Replicate's cache when identical inputs would return the same image.
How to install it
One install covers every Replicate node.
- ComfyUI Manager: search ComfyUI-Replicate, install, restart.
- Manual:
then restart.cd ComfyUI/custom_nodes git clone https://github.com/replicate/comfyui-replicate cd comfyui-replicate pip install -r requirements.txt
No weights download - 1.1 Pro is closed and runs on Replicate. Only the replicate pip package installs. Set your token first: export REPLICATE_API_TOKEN="r8_..." (replicate.com/account/api-tokens).
Common issues & troubleshooting
Where are the steps and guidance sliders? Gone on purpose. 1.1 Pro is tuned to a fixed high-quality pipeline, so it doesn't expose them. If you specifically want to tune steps and guidance, use flux-pro (the original) or flux-dev in this same pack - but for most people the tuned defaults here are the point.
My custom width/height did nothing. They only apply when aspect_ratio is custom. Set the ratio to custom first, then width/height take effect.
prompt_upsampling changed my prompt. That's its whole job - it rewrites your prompt for more detail. Turn it off if the exact wording (or spelling of any text) matters.
No negative prompt. Correct, and it's Flux-wide. Steer with the positive prompt only. If a run returns the same image, the Replicate cache fired for identical inputs - change the seed or set force_rerun.
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | — | |
| aspect_ratioopt | COMBO | 1:1 | 10 options: custom, 1:1, 16:9, 2:3, 3:2, 4:5, +4 |
| widthopt | INT | 256–1440 | — |
| heightopt | INT | 256–1440 | — |
| safety_toleranceopt | INT | 21–5 | — |
| seedopt | INT | — | |
| prompt_upsamplingopt | BOOLEAN | false | — |
| output_formatopt | COMBO | webp | 3 options: webp, jpg, png |
| output_qualityopt | INT | 800–100 | — |
| force_rerunopt | BOOLEAN | false | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| IMAGE | IMAGE | — |