Lora Queue Helper

Lora Queue Helper is the batch twin of a plain LoraLoader. Where LoraLoader applies one LoRA and hands you one model/clip pair, this node takes a list of LoRA names and hands you a list of model/clip pairs - one per entry, all at a single strength. Feed the outputs to a batch-aware sampler and you get one image per LoRA in a single queue run.

That's the workflow it's built for: pick a checkpoint, drop a bunch of LoRAs you want to audition into lora_list (one filename per line), set one strength, and let it fan out. Pair it with the pack's Lora List Helper - which outputs exactly the newline-separated format this node wants - and you can literally drag your entire LoRA folder into a queue. For anyone who's ever sat there manually loading and unloading LoRAs to compare them, this is the cure. It also pairs with the LoRA Grid nodes if you want labels and a comparison sheet instead of a raw batch.

How it works. Each line in lora_list is resolved through ComfyUI's LoRA folder lookup, loaded, and applied to your input model and clip at the given strength. Three outputs come back as parallel lists: MODEL, CLIP, and lora_info (a human-readable name @ strength string per entry). If a name isn't found, it doesn't crash - that entry passes the original model/clip through and marks lora_info with [NOT FOUND], so you can spot the typo without the whole run dying. Same graceful behavior on load errors.

Inputs and outputs. Required: model (MODEL), clip (CLIP), lora_list (multiline STRING), strength (FLOAT, −10 to 10, default 1.0). Outputs are the three lists above. Note the outputs are list-typed, so a downstream node has to accept lists - check that your sampler does, or you'll be scratching your head.

Install.

cd ComfyUI/custom_nodes
git clone https://github.com/tetsuoo-online/Comfyui-TOO-Pack

or ComfyUI Manager → "Comfyui-TOO-Pack" → restart. No dependencies beyond what ComfyUI ships.

Where people get burned. The strength field applies to every LoRA in the list - there's no per-line strength here. If you want each entry at its own weight, that's what the LoRA Grid nodes are for (they parse name:0.7 style entries). And lora_list wants filenames as ComfyUI knows them (my_style.safetensors), not full paths. Keep the list to the same LoRA architecture as your checkpoint, obviously - an SDXL LoRA queued on a Flux model does nothing useful, which is the same architecture-bound rule that applies everywhere (see the LoRA concept notes). For auditioning a folder of candidates in one go, it's hard to beat.