To my recollection i found some small differences between crees/samsungs 2700k 90s and bridgelux of same spectrum. Iirc correct from talk around threads and those old threads of Supra and Cobkits where they looked at cobs with licor: bridgelux tends to have slightly redder spectrums with less green but a little more cyan. Ive done both crees and blux in that spectrum but not side by side so hard to say but my guess is that they push a bit more red. All cct arent created equal but differences are still minimal.
Be careful about ai, they do not reveal truth, they reveal truthy. Job wel done is you giving them thumbs up, happy with the answer. My dad does family research going back long time. I once asked ai about a name that surfaced and got the response that he was the foreign minister of my country over several governments including over ww2! I was ecstatic and proud but double checked and it was all a lie; i guess ai thought i could use a little boost and made something up.
Ive also had very dubious behaviour from ai when i started to ask it questions it didnt like, almost passive aggressive and going from responding in a way i "trained" it back to responding in the most impossible to understand way it could find.
I think its just good for creating some blurby content and to use as a sort of google for complicated questions; when standard search terms wont do it. Dont try to get language models to do math for you. You have to check every single step of what it does, and it will never give any indication on when its actually doing things right or wrong.