Yes, we have an enclosure, but that wasn't a problem so far. I'll add it to the things to try. Thank you. :)

I'm using manufacturer (model designer and filament manufacturer) provided files. If I remember correctly, I tried a lower speed, but I will experiment more with the speed since it was suggested so often. I think, it should be a constant speed, since it's all external perimeter and looks to be set to a constant value.

It's not vase mode, but the model creator designed it to print layers without retraction. It only does a retraction to start the next layer.

Last time I tried, I think, I did multiple attempts back-to-back and starting again at layer 1 was no problem, so I don't think it was clogged. I'll experiment more with temperature and speed like you and others suggested.

The layer doesn't look unusual compared to the previous layers, but thanks for the suggestion.

Do you mean the exact same model but another filament like regular PLA? Other models with regular PLA is no problem.

Other commenters also suggested to experiment with temperature. We'll also do some speed tests. Thanks.

It's LW-PLA, that foams, could be necessary for the foaming. We just used the filament by the manufacturer with their model files and presets.

Good idea. I would have expected new filament to be dry. If nobody else comes along and says: It's definitely 'this', we'll try to dry it. Thanks.

I think the issue is how the games are programmed. Some games check which controllers are plugged in only at start and some do regular checks during runtime.

From the top of my head: the context size is way higher. 128k tokens vs 8k usually.

You're not alone in that. I remember when Blender 2.80 was about to release and it was a big milestone and everybody online was hyping about "Blender 2.8". I think even Blender themselves went with the flow and changed the marketing.