Reference only

LoRA has no effect or reports size mismatch: check the base model and both output branches

A selectable LoRA can still target an incompatible model, or the workflow can bypass its modified outputs. Little visible change is an observation, not proof of a damaged file.

Symptoms and scope

A selectable LoRA can still target an incompatible model, or the workflow can bypass its modified outputs. Little visible change is an observation, not proof of a damaged file.

Error fragments for search; IDs, values and filenames may vary:

lora key not loaded
LoRA
size mismatch

Source-supported context

Official LoRA examples route model and CLIP through the loader and distinguish strength_model from strength_clip. Component-family compatibility should be checked before increasing strength. ComfyUI logs lora key not loaded for keys left unmapped by its loader; this warning alone does not identify a single cause or prove that every layer failed. Source 1 Source 2 Source 3

Cases to distinguish

1. The LoRA targets another base family or variant with different layers and dimensions.

2. Sampling or text encoding still uses an earlier model/CLIP branch, bypassing the loader outputs.

3. Strength and prompt conditions do not match the intended effect; some weights do not modify CLIP at all.

Diagnostic sequence

The sequence below is an editorial procedure based on the cited context, not a diagnosis already confirmed for your environment.

Step 1. Record the author's base model, variant and usage instructions. Mark compatibility unknown when the source does not state it.

Step 2. Follow both model and text-encoding outputs to confirm the actual sampling path uses the modified objects.

Step 3. Compare with/without the LoRA using the same seed, base and dimensions, starting within the author's suggested range and without stacked LoRAs.

Step 4. Preserve key/shape warnings and check whether partial-layer loading was expected. Route severe dimension errors to component compatibility diagnostics.

Completion check

The workflow uses the intended weights and the logs and controlled comparison are explainable. Record subjective quality separately from technical loading success.

Limitations and cautions

A filename does not prove compatibility with every SDXL, FLUX or similarly named derivative. No commercial-use permission is inferred for third-party weights.

Original sources

Sources and loader code rechecked on 2026-09-25. The 2026-09-21 dates above record the earlier source review. No GPU run, minimum-memory measurement or runtime guarantee is supplied.

Didn't solve it?

Check the next possible cause

The same symptom can come from a different cause. Work through these related entries in order.

  1. mat1 and mat2 shapes cannot be multiplied: inspect the model component chainmat1 and mat2 shapes cannot be multipliedMatrix dimensions help locate a failure during sampling, encoding or adapter use, but one numeric pair does not identify a unique missing model.
  2. Value not in list: why a downloaded model does not appear in ComfyUIValue not in listA saved ckpt_name, lora_name or another selection is absent from the current allowed list. Check the input name first: this validation error can also concern non-file options such as samplers.
Paste your full log into the error finder

Was this page helpful?

Anonymous. Only a yes/no counter is stored; no account, IP address or device details.

Sources & references

Official LoRA documentation, model troubleshooting and the v0.37.0 loader code were rechecked on 2026-09-25. No GPU run, minimum-memory measurement or runtime guarantee is supplied.

01LoRA workflowSource checked: 2026-09-2502ComfyUI model troubleshootingSource checked: 2026-09-2503ComfyUI v0.37.0 LoRA key loaderSource checked: 2026-09-25

Report an issue · 2934af59-c879-5b50-851d-9734917a082a