AI300
  1. alphaXiv
    2041.
    "First Proof" A team of researchers proposes a way to test if AI can actually do NEW math by releasing 10 freshly-solved and never public research questions, with answers temporarily encrypted.›
    LLM score 92 · 6 months ago
  2. Boris Cherny / Creator of Claude Code
    2042.
    We just launched an experimental new fast mode for Opus 4.6.›
    LLM score 85 · 6 months ago
  3. Jerry Tworek / ex OpenAI VP of RL
    2043.
    Run fewer experiments and think about them more
    LLM score 85 · 6 months ago
  4. Cameron Wolfe / Researcher at Netflix
    2044.
    Still can't believe Olmo 3 found out you can do in-flight weight updates during RL without even clearing the KV cache and it still works.›
    LLM score 92 · 6 months ago
  5. alphaXiv
    2045.
    "Maximum Likelihood Reinforcement Learning" Turns out we've been using RL wrongly the whole time, and missed out on 20x efficiency?›
    LLM score 92 · 6 months ago
  6. Yifei Zhou / Thinking Machines Researcher
    2046.
    Always a big fan of works coming out of Andrea’s lab! Love the mathematical insight from the comparison of rl and maximum likelihood loss https://t.co/ZhaYCGyNwF
    LLM score 12 · 6 months ago
  7. Omar Khattab / MIT CSAIL Asst professor
    2047.
    wait that’s such a smart product move from OpenRouter›
    LLM score 68 · 6 months ago
  8. Igor Babuschkin / Cofounder of xAI
    2048.
    I’ve tested the latest generation of all the major AIs on theoretical physics research and Claude 4.6 has absolutely blown me away with how capable it is in physics.›
    LLM score 92 · 6 months ago
  9. Omar Khattab / MIT CSAIL Asst professor
    2049.
    Since there's now 5+ papers that propose on-policy context distillation, I feel comfortable confessing that we (@NoahZiems), too, were working on that haha.*›
    LLM score 92 · 6 months ago
  10. Cyris Kissane / Researcher at Flapping Airplanes
    2050.
    Cool new paper: Drifting Models. https://t.co/ALvy7dmbWC›
    LLM score 75 · 6 months ago
  11. Cyris Kissane / Researcher at Flapping Airplanes
    2051.
    https://t.co/zcejLXVsTf cool new secure python interpreter for AI use.›
    LLM score 75 · 6 months ago
  12. Ben Burtenshaw / Hugging Face Researcher
    2052.
    the @huggingface supports agentic environments that you can use for rl or evals.›
    LLM score 85 · 6 months ago
  13. John Carmack
    2053.
    256 Tb/s data rates over 200 km distance have been demonstrated on single mode fiber optic, which works out to 32 GB of data in flight, “stored” in the fiber, with 32 TB/s bandwidth.›
    LLM score 92 · 6 months ago
  14. Cameron Wolfe / Researcher at Netflix
    2054.
    The Elon / Dwarkesh podcast was cool.›
    LLM score 82 · 6 months ago
  15. alphaXiv
    2055.
    New generative model paradigm: Drifting Models 1 step inference but at SoTA fidelity!›
    LLM score 92 · 6 months ago
  16. Jeff Dean / Chief Scientist at DeepMind
    2056.
    Gemini+Genie 3 are helping @Waymo simulate long tail scenarios to make driving safer.›
    LLM score 75 · 6 months ago
  17. Sharon Zhou / AMD VP of AI
    2057.
    I like to think of evals as something active, not passive -- it's a North Star that steers LLMs toward higher intelligence.›
    LLM score 92 · 6 months ago
  18. Yingru Li / xAI Researcher
    2058.
    We trained Dr. Kernel - a 14B model that matches or beats GPT-5 and Claude-4.5-Sonnet on KernelBench.›
    LLM score 92 · 6 months ago
  19. Ben Burtenshaw / Hugging Face Researcher
    2059.
    Eval scores in 2026 are broken.›
    LLM score 92 · 6 months ago
  20. Merve Noyan / Hugging Face ML Engineer
    2060.
    we released Community Evals to fix transparency in evals 🤝›
    LLM score 85 · 6 months ago
  21. Thomas Wolf / Hugging Face Cofounder
    2061.
    [On AI lying] Convergence of reading in my list today between Anthropic's fresh Opus 4.6 model card and @dwarkesh_sp's interview of Elon on the question of training powerful AI model to/on lies:›
    LLM score 75 · 6 months ago
  22. Cameron Wolfe / Researcher at Netflix
    2062.
    This is a really nice framing of efficiency constraints in RL training.›
    LLM score 92 · 6 months ago
  23. Ted Sanders / OpenAI Researcher
    2063.
    it's wild how much LLMs have transformed the profession of software engineering over the past year.›
    LLM score 85 · 6 months ago
  24. Kevin Murphy / DeepMind Research Scientist
    2064.
    I'm pleased to share our paper on learning temporally abstract world models and policies (options).›
    LLM score 95 · 6 months ago
  25. Sholto Douglas / Researcher at Anthropic
    2065.
    Default case right now is a software only singularity, we need to scale robots and automated labs dramatically in 28/29, or the physical world will fall far behind the digital one - and the US won’t be competitive unless we put in the investment now (fab, solar panel, actuator›
    LLM score 92 · 6 months ago
  26. Richard Song / DeepMind Research Scientist
    2066.
    “Am I disappointed when my students don’t go off and be mathematicians?›
    LLM score 85 · 6 months ago
  27. Sholto Douglas / Researcher at Anthropic
    2067.
    Long weekends are good for GDP because people discover how good the models are https://t.co/yucnlrRos4
    LLM score 85 · 6 months ago
  28. Sherwin Wu / OpenAI API, Head of Engineering
    2068.
    This article really resonated with me, especially in relation to GPT-5.3-Codex.›
    LLM score 85 · 6 months ago
  29. Geoffrey Hinton
    2069.
    This is a great report that provides a thoughtful, detailed and very well researched description of the risks of AI.›
    LLM score 92 · 6 months ago
  30. Andrew White / Edison Scientific Cofounder
    2070.
    OpenAI Ginkgo collab is today, Opus 4.6 is today, 5.3-codex is today, METR 5.2-codex was last night, and we decided to do labbench2 today.›
    LLM score 92 · 6 months ago
More