Documents are processed as tokens as well, unless its bitmap is ocr'd.
Images tho are natively compatible with Multi-Modal LLMs, so theres no image->text translation layer in between.
It's that the unit of cost is different (e.g. "visual token" vs text token)
But the same article has the following quote from the above video, from Bayer's executive Stefan Oelrich:
> “Ultimately the mRNA vaccines are an example for that cell and gene therapy. I always like to say, if we had surveyed two years ago in the public, ‘would you be willing to take gene or cell therapy and inject it into your body?,’ we would have probably had a 95% refusal rate. I think this pandemic has also opened many people’s eyes to innovation in the way that was maybe not possible before.”
So it seems Bayer (or at least, Stefan Oelrich from Bayer), indeed classifies the mRNA vaccines as gene therapy.
Gah - this is unfortunate. This will have me quit as well - I wonder when they will begin to enforce this on the Claude Code front. I remember when OpenAI tried (maybe they still are?) putting this in for GPT 5 and this had me switch from them to Anthropic.
Aman premoves everything in the end, he is very low on time. Most platforms allow you to make moves in advance, and those are automatically executed after your opponent moves, as long as they are legal.
RedHat one is buggy and slow, but it works after tinkering and display a lot of warnings and code suggestions.
This extension loads instantly without any issues but it show 0 warnings/errors in my project. "Find all references" works as expected.
Would that reduce the number of tokens used too?