Comment by bel8
14 hours ago
I think they had this sitting idle for a while, waiting for their AI division to deliver a capable LLM to power it.
Now they have Muse Spark 1.3, which is somewhere between Sonnet and Flable (Opus tier in my experience) in intelligence, but is very fast at 200+ tok/s.
I have been using Muse Spark 1.3 contrib (the one that shares your data for training but is much cheaper) via OpenCode Go $10/mo plan for coding and the thing flies, it's a pleasure to use. I think the only faster acessible model currently is Gemini 3.8 Flash which is around 250-280 tok/s.
The amount of usage you get is insane when you stack OpenCode GO subsidy on top of contributor.
There is already evidence of groups setting websites and "think tanks" that aren't for human but for influencing AIs scraping them.
I wonder if having the contributor model read and summarize your propaganda would be even more effective. Instead of astroturfing reddit, just feed it to the model directly. Give it generated threads and ask it loaded questions.
Poisoning LLM? I remember at least one study about it. Perhaps but it would have to be very subtle otherwise they might flag your content and shadowban it from future training.
My personal opinion is that trying to make LLMs inefficient is futile and causes even more waste of earth resources.
It's like trying to stop a coal-powered locomotive by parking a truck on the rails. You're just going to make the thing burn more coal before reaching it's destination.
I wasn't talking about poisoning LLM for generally degrading performance/improvement. But for manipulating answers for propaganda/ideological reasons.