AI300
  1. Noam Brown / OpenAI Research Scientist
    91.
    I vibecoded an open-source poker river solver over the holiday break. The code is 100% written by Codex, and I also made a version with Claude Code to compare. Overall these...
    LLM score 75 · 24 days ago
  2. Igor Babuschkin / Cofounder of xAI
    92.
    @mmt_lvt I personally find Scala too complex. It’s giving me C++ vibes in the sense that there are many advanced language features which become footguns when used in practice due...
    LLM score 20 · 24 days ago
  3. Igor Babuschkin / Cofounder of xAI
    93.
    @y0b1byte I haven’t encountered any issues with writing proc macros so far with Opus. Just asked it to write a decently complex one 30 minutes ago
    LLM score 20 · 24 days ago
  4. Igor Babuschkin / Cofounder of xAI
    94.
    Opus 4.5 is surprisingly good at writing decent Rust code and I expect this will make the language much more popular than it already is. Rust used to have a steep initial learning...
    LLM score 40 · 24 days ago
  5. Igor Babuschkin / Cofounder of xAI
    95.
    @DavidSHolz There are decades where nothing happens, and there are weeks where decades happen
    LLM score 20 · 26 days ago
  6. Andrej Karpathy / AI researcher
    96.
    The first 100% autonomous coast-to-coast drive on Tesla FSD V14.2! 2 days 20 hours, 2732 miles, zero interventions. This one is special because the coast-to-coast drive was a...
    LLM score 20 · 29 days ago
  7. Andrej Karpathy / AI researcher
    97.
    @eiselems Claude has been running my nanochat experiments since morning. It writes implementations, debugs them with toy examples, writes tests and makes them fail/pass, launches...
    LLM score 70 · about 1 month ago
  8. Tim Dettmers / Research Scientist at Ai2
    98.
    @HeMuyu0327 I literally wrote a paper about this. Attention sinks = large outliers in LLM.int8(). The dimensions selected increase with scale/perplexity: https://t.co/eNpinYaadd
    LLM score 80 · about 1 month ago
  9. Demis Hassabis / CEO of DeepMind
    99.
    ‘The Thinking Game’ documentary has just passed 200M views on YouTube in just 4 weeks! 🤯Perfect holiday viewing if you’re interested in a behind-the-scenes look at how an AGI lab...
    LLM score 20 · about 1 month ago
  10. Jim Fan / NVIDIA Director of Robotics
    100.
    Everyone's freaking out about vibe coding. In the holiday spirit, allow me to share my anxiety on the wild west of robotics. 3 lessons I learned in 2025. 1. Hardware is ahead of...
    LLM score 20 · about 1 month ago
  11. Andrej Karpathy / AI researcher
    101.
    @shazow Very good questions imo experienced devs have a real advantage but only if they rapidly progress through their grief cycle and adapt, now and onwards. Categorically...
    LLM score 60 · about 1 month ago
  12. Jim Fan / NVIDIA Director of Robotics
    102.
    2024: AI is the copilot 2025+: humans are the copilot Copilot is the new engineering skill. It’s not easy to leave the driver seat - we must learn to think the AI way and adapt to...
    LLM score 60 · about 1 month ago
  13. Andrej Karpathy / AI researcher
    103.
    @ibab It’s very good. People who aren’t keeping up even over the last 30 days already have a deprecated world view on this topic.
    LLM score 20 · about 1 month ago
  14. Andrej Karpathy / AI researcher
    104.
    I've never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I...
    LLM score 75 · about 1 month ago
  15. Andrej Karpathy / AI researcher
    105.
    @bcherny Ran into token file context limits this morning. It's possible to override them for the MCP tool setting MAX_MCP_OUTPUT_TOKENS but I don't believe an equivalent exists...
    LLM score 70 · about 1 month ago
  16. Jonathan Ross / TPU Creator
    106.
    Today Groq entered into a non-exclusive licensing agreement with Nvidia for Groq’s inference technology. Along with other members of the Groq team, I’ll be joining Nvidia to help...
    LLM score 50 · about 1 month ago
  17. Andrej Karpathy / AI researcher
    107.
    @ahall_research fun! I should mention that my original council was different LLMs but I think it should work well (possibly even better?) even with one LLM but different prompts....
    LLM score 60 · about 1 month ago
  18. Jim Fan / NVIDIA Director of Robotics
    108.
    Visual sim2real: zero-shot deploy to the real world, with zero real data. Trained entirely in Isaac Lab. https://t.co/aRuvwi34oi
    LLM score 80 · about 1 month ago
  19. Jim Fan / NVIDIA Director of Robotics
    109.
    FLARE: a latent world model technique that boosts GR00T N1.5 significantly. https://t.co/Kk0gjxDoRq
    LLM score 85 · about 1 month ago
  20. Jim Fan / NVIDIA Director of Robotics
    110.
    RL can post-train VLA models to close to 100% robustness on high-precision tasks like GPU insertion. It's a major step towards the last-mile problem of industrial...
    LLM score 85 · about 1 month ago
  21. Jim Fan / NVIDIA Director of Robotics
    111.
    SONIC: a 40M-parameter foundation model that learns the subconscious, whole-body motor coordination skills. https://t.co/SnkXPLTRHS
    LLM score 85 · about 1 month ago
  22. John Schulman
    112.
    Humans are jagged, and organizations (from companies to civilizations) have evolved as harnesses to make best use of us despite our faults https://t.co/rJFqEMBj5I
    LLM score 20 · about 1 month ago
  23. Igor Babuschkin / Cofounder of xAI
    113.
    @demishassabis Solomonoff induction suggests that an asymptotically optimal general learning machine exists, at least theoretically.
    LLM score 85 · about 1 month ago
  24. Demis Hassabis / CEO of DeepMind
    114.
    Yann is just plain incorrect here, he’s confusing general intelligence with universal intelligence. Brains are the most exquis​ite and complex phenomena we know of in the...
    LLM score 75 · about 1 month ago
  25. Andrej Karpathy / AI researcher
    115.
    https://t.co/Lb6T42n5jl
    LLM score 90 · about 1 month ago
  26. Demis Hassabis / CEO of DeepMind
    116.
    @GoogleDeepMind has led the way in applying AI to science. Great to see the White House’s new Genesis Mission & excited to be collaborating with @ENERGY by giving National Lab...
    LLM score 75 · about 1 month ago
  27. Tri Dao / Chief Scientist at Together
    117.
    This is what we've been coking for the last 9 months: make MoEs training goes ~2x faster and ~2x less memory! Highlights: - MoE typically takes the most time and memory in modern...
    LLM score 85 · about 1 month ago
  28. Noam Shazeer
    118.
    Gemini 3 Flash is live. ⚡️ We’ve packed Gemini 3’s Pro-grade reasoning into a leaner model with Flash-level latency, efficiency, and cost. It's my favorite model to use – the...
    LLM score 70 · about 1 month ago
  29. Demis Hassabis / CEO of DeepMind
    119.
    For a fast model, Gemini 3 Flash offers incredible performance, allowing us to provide frontier intelligence to everyone globally. Try the 'fast' mode from the model picker in the...
    LLM score 20 · about 1 month ago
  30. Noam Brown / OpenAI Research Scientist
    120.
    Our efforts at @OpenAI to advance scientific progress aren't just limited to math/physics/coding. In this work, GPT-5 was evaluated on its ability to optimize wet lab experiments....
    LLM score 20 · about 1 month ago
More