@ns123abc
I remember when DeepMind CEO called the DeepSeek hype exaggerated but nobody wanted to hear it because the narrative was Google was falling behind Demis Hassabis in 2025: >"a lot of the claims are exaggerated and misleading" >"they report the final training run — a fraction of what it takes to explore and train" >"they rely on Western models to distill from" >"no new silver bullet technologies we haven't seen or invented before — they just applied it well" >"gemini is more efficient" 14 months later: >Gemini 2.5 Pro is frontier-tier >V4 has been delayed for months with no explanation >Deepseek is raising $300 million after saying "money was never the problem" The "self-funded research lab" narrative is officially over