Mastodon Feed
Boosted by kornel ("Kornel"):
mattmay@mstdn.social ("Matt May") wrote:
RE: https://mastodon.social/@verge/117259017019721108
It is far better to read this as an attempt to block DeepSeek and the other cheaper-but-still-capable models from the US market, allowing OpenAI/Anthropic to cut way back on the model training that costs them billions.
They want to do to LLMs what their own behavior did to RAM/GPUs: raise their prices not because they're any better or faster, but by forming a cartel that controls production so they can charge whatever they like.