Log inSign up
Hexiang (Frank) Hu
1,006 posts
Hexiang (Frank) Hu profile banner
@hexiang

Hexiang (Frank) Hu

@hexiang
@SpaceXAI: @grok multimodal, imagine | prev @GoogleDeepMind: gemini & imagen
Earth
hexiang-hu.github.io
Joined September 2014
554
Following
5,936
Followers
1
Subscription
RepliesRepliesRepostsRepostsMediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
  • Pinned
    @hexiang
    Hexiang (Frank) Hu
    @hexiang
    Aug 14
    Thrilled to receive this feedback from Tesla — they tackle the world’s most critical multimodal challenges. Extremely proud that the team’s work is paying off🚀
    @yunta_tsai
    Yun-Ta Tsai
    @yunta_tsai
    Aug 14
    Grok 4.6 multimodal is a step change from Grok 4.5. It’s one of the under-discussed improvements, and I’ve been very impressed by it. My daily work includes reviewing lots of videos and understanding the context; Grok 4.6 improves the productivity of such workloads by at least
    52
  • @hexiang
    Hexiang (Frank) Hu
    @hexiang
    Sep 8
    True
    @_arohan_
    rohan anil
    Core Automation
    @_arohan_
    Sep 8
    Whatever people say of Gemini and they will rebound. Google probably has the most hardened privacy infra for data of all time. It's extremely well protected.
  • @hexiang
    Hexiang (Frank) Hu
    @hexiang
    Sep 5
    LFG
    @YknZhu
    Yukun
    @YknZhu
    Sep 5
    Image + video + smart agents work better together at grok.com/imagine/agent A solid step in agentic video creation — we’d love your feedback!
  • @hexiang
    Hexiang (Frank) Hu
    @hexiang
    Sep 3
    Looped transformers aren’t a new species. They’re a way to buy FLOPs without buying parameters. If intelligence tracks useful compute more than weight count, that’s the correct trade.
    @savvyRL
    Rosanne Liu
    @savvyRL
    Sep 2
    Stop blowing up Looped Transformers — it's not magical, it doesn't add recurrence to the (still and always) feedforward network, it was done years ago (Dehghani et al., 2019). All it does is making your model deeper, the same as if you added more layers to begin with.
    1
  • @hexiang
    Hexiang (Frank) Hu
    @hexiang
    Aug 20
    Imagine 2 is designed for real-world usefulness. This is it working in the hands of art professionals — assets rebuilt from screenshots, already usable. Thanks Dogan for sharing this feedback over♥️
    @doganuraldesign
    Dogan Ural
    @doganuraldesign
    Aug 20
    Just got this text from an old friend. His exact words: "Grok Imagine turned out really good, man. It saved us from ChatGPT. I gave it two screenshots from our game, and it went through them one by one, recreated all the items, cropped them, and gave me EVERYTHING
    5
Advertisement
Advertisement