top 50 comments

sorted by: hot top controversial new old
[–] [S] 70 points 1 week ago* (last edited 1 week ago) (3 children)

Just a day ago, a senior Anthropic executive claimed the U.S. still held a 6 to 9 month lead in frontier AI models, while calling Chinese model distillation adversarial. One day later, Moonshot’s Kimi K3 beat Claude Fable 5 on Frontend Code Arena. And the funniest part is that it is an open weight model.

So the whole 6 to9 month lead lasted about 24 hours. 🤣

https://hai.stanford.edu/news/inside-the-ai-index-12-takeaways-from-the-2026-report

  • source
  • hideshow 5 child comments
  • [–] 9 points 1 week ago (1 child)

    Tbf they (Moonshot) claim it’s not as good as Fable overall. Fable is also held back by US government mandated guards. Anthropic could well be ahead by a lot in terms of research and maybe we just don’t have the full picture. But that puts them in an even worse spot because it would mean they’re already at the limit of what they can compete with in the market.

  • source
  • parent
  • hideshow 2 child comments
  • [–] [S] 17 points 1 week ago (2 children)

    I think the big picture here is that the difference in quality is largely subjective at this point, while US companies are burning through orders of magnitude of cash which is obviously not sustainable. We shouldn't underestimate the power of developing things in the open. Chinese open models benefit from the wisdom of an entire global research community while American engineers working on proprietary closed models are working in their own insular silos. It should be no surprise that the scientific community at large would pull ahead of these small teams. On top of that, doing research in the open amortizes the cost. Incidentally, this is exactly the same logic that led open source to dominate in recent years.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 4 points 1 week ago (3 children)

    OpenAIs bet was that compute was the limiting factor. Sam Altman claimed that essentially no-one could obtain the level of compute necessary to develop something like GPT-4. Then Deepseek came out.

    They're still trying to make the original claim true because if they admit that they were wrong (or full of shit) it all comes crashing down, not just their companies but likely the economy as a whole.

  • source
  • parent
  • hideshow 4 child comments
  • load more comments (2 replies)
  • [–] 3 points 1 week ago

    It doesn’t hurt that American frontier models are phenomenally huge and cost an exorbitant amount per token in and out while the main source of real world value created by llms in the past year has been through agentic/harness/other words for massive context systems that show a real benefit from simply using more tokens.

  • source
  • parent
  • load more comments (1 reply)
    [–] 36 points 1 week ago* (last edited 1 week ago) (1 child)

    Why are there so many pictures of Sam Altman just Looking Like That?

  • source
  • hideshow 2 child comments
  • [–] 19 points 1 week ago (82 children)

    First bomb they've dropped in 50 years, wow.

  • source
  • hideshow 82 child comments
  • load more comments (82 replies)
    [–] 7 points 1 week ago (3 children)

    The latest model is K2.7, I can decide what wrote the article. Is it more likely than an Ai did not know 2.7 existed or a journalist at Gizmodo is incompetent?

  • source
  • hideshow 4 child comments
  • load more comments (2 replies)
    [–] 7 points 1 week ago
    [–] 5 points 1 week ago (4 children)

    Can we please not write headlines like that when we're on the verge of ww3 and literal bombs might be dropped on the US in the coming months?

  • source
  • hideshow 4 child comments
  • load more comments (4 replies)
    [–] 5 points 1 week ago (8 children)

    Kimi K3 looks insane but it's also very expensive right now, hopefully that changes or I don't see myself using it over the previous models

  • source
  • hideshow 9 child comments
  • load more comments (7 replies)
    load more comments
    view more: next ›