Flux Delete Finetune (BFL)

Flux Delete Finetune (BFL) is exactly what it sounds like: it deletes a finetune from your BFL account. You hand it a finetune_id, it calls BFL's delete_finetune endpoint, and the custom model is gone. It's the last node in the ComfyUI-FLUX-BFL-API pack's finetune-management loop - My Finetunes to list, Details to inspect, Status to track, Delete to retire.

You probably won't use this one often. But when you've got stale test finetunes piling up on a paid account - or you trained something you don't want on your account anymore - this is the node that saves you a trip to the web dashboard. It's also a good way to keep your account tidy if BFL's finetune storage is metered, which hosted finetunes usually are.

One thing to be completely clear about: this is destructive and immediate. There's no confirmation step beyond the node's own execution, and BFL isn't going to keep a copy because you changed your mind. Double-check the ID before you queue.

How it works

It's a POST to the delete_finetune endpoint with the finetune_id in the body. The response comes back as JSON through the STRING output - on success, the API's confirmation payload; on failure, the error. The node prints what it's doing to the ComfyUI console (🗑️ Deleting finetune ID: ...) so you can see it happen in the logs.

The inputs and output

• finetune_id - the ID of the finetune to delete. The only input that matters, and the one to be sure about.
• config (optional) - a Flux Config (BFL) node. The regional rule applies here too: region must be us or eu because the default api.bfl.ai/v1/ doesn't serve finetune routes.
• Output: STRING - the JSON confirmation or error.

Installing it

Standard pack install:

cd ComfyUI/custom_nodes
git clone https://github.com/gelasdev/ComfyUI-FLUX-BFL-API.git

Restart, key set, region set on your config node. torch is the only declared dependency - nothing else to install.

Where people get burned

• Deleting the wrong model. There is no undo. If you're not certain an ID maps to the finetune you think it does, run Flux Finetune Details on it first. Thirty seconds of checking beats a retrained model.
• Region not set. The recurring finetune-section failure: no region on the config node means the call never reaches the right endpoint and returns an error (which, at least here, means your finetune survives - the silver lining of the error path).
• Expecting the finetune to be gone from local tools too. This deletes it from BFL's account. Any workflow referencing that ID will now fail at generation time - clean up your workflows when you clean up the model.

Delete nodes are never glamorous, but this one completes the finetune management story the pack tells: train, list, inspect, track, and eventually retire. Keep it wired into your account-tidying workflow and you'll never touch the BFL dashboard again.