No scorecards, no winner declared in the first paragraph. Each page lays out the trade-off, says plainly when each option is the right one, and tells you what to measure on your own data.
Same adapter maths. Very different memory bills.
LoRA and QLoRA both freeze the base model and train a small set of low-rank adapter matrices. The difference is what precision the frozen base is held at during training — and that single choice decides which GPU you need.
One teaches behaviour. The other supplies facts.
These are not competing answers to the same question. Fine-tuning changes how a model responds; retrieval changes what it knows at the moment you ask. Most production systems that work well use both, for different reasons.
Who holds the weights decides who holds the leverage.
Serving an open-weights model you control and calling a hosted proprietary API are different operating models, not just different price points. The trade-off is convenience against control — over data, versions, cost curve and exit.
Update every weight, or a carefully chosen few.
Full fine-tuning updates every parameter in the model. Parameter-efficient fine-tuning (PEFT) freezes the base and trains a small set of added parameters instead. The gap between them is narrower than most teams expect, and the operational difference is large.