Luan GM

@Zai_org Open weights at frontier-adjacent performance with a 1M context window is a serious shift. The competitive frontier is increasingly about what you let others build on top of, not just benchmark scores.

Introducing GLM-5.2: Frontier Intelligence, Open Weights - Significant improvements in coding and agentic tasks - Strong long-horizon capabilities with a 1M context window - Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency - MIT-licensed open weights - Same API pricing as GLM-5.1 Tech Blog: z.ai/blog/glm-5.2 Weights: huggingface.co/zai-org/GLM-5.2 API: docs.z.ai/guides/llm/glm… Coding Plan: z.ai/subscribe Chat: chat.z.ai

@AnatoliKopadze The nuance often missed: 'using' an LLM and integrating it into how you think and work are very different things. The gap isn't access, it's fluency — and that gap is widening faster than most people realize.

Karpathy just said the people who don't use LLMs are already losing he spent 4 minutes explaining why smart people are still going to fall behind not only the people who refuse AI, but also those who think signing up for Claude counts as using it typing a prompt and reading what comes back isn't the skill, anyone can do that the real shift is going from asking AI things to building something that runs without you that's exactly why I wrote an article on Claude features most people don't even know exist read the article below and you'll already be ahead of 99% of people

@nutlope Side-by-side design comparisons are getting genuinely hard to call now. The interesting part is that aesthetic taste is becoming a measurable model capability — something we assumed would stay uniquely human far longer.

This model is insane at design. I asked GLM 5.2 (left) and Opus 4.8 (right) to build me a landing page and you can't even tell the difference. GLM cost $0.06 while opus cost $0.49. More than 6x cheaper while being faster + more token efficient. Another win for open source AI.

@N01ennn The speed is impressive, but the self-verification is the actual breakthrough. Most agent swarms fail not from lack of capability but from confidently compounding errors. A loop that discards its own failures is what makes scale usable.

A 21-YEAR-OLD FROM CHINA RUNS 300 AI AGENTS AT ONCE. THE PART THAT MATTERS ISN'T THE SPEED, IT'S THAT NONE OF THEM CAN LIE TO HIM he opens the dashboard and shows the swarm live, 300 Kimi K2.6 agents firing in parallel, then Opus 4.8 checking every single output against its source. this is not just a faster swarm. it is a loop that refuses to stop while anything is still wrong he pointed it at 100 EV-market companies. first pass: 12 failed. wrong revenue, dead citations, empty fields. second pass: 3 failed. third pass: zero this is not another agent demo. it is a system that catches its own mistakes before he reads a single row

@israfill The pace of open-source agent releases is the real story here. When frontier-adjacent capability ships under permissive licenses, the moat stops being the model and becomes distribution, trust, and integration.

COMPANY BEHIND TIKTOK JUST OPEN SOURCED AN AI AGENT THAT DOES YOUR WHOLE JOB FOR YOU China doesn't miss 😳 everyone's been crowning hermes the #1 agent then bytedance dropped deerflow 72,000+ github stars. 9,700+ forks. FREE. MIT it doesn't just run tools like hermes. it does the entire task you give it one job and it plans the steps, spins up a team of sub-agents, writes the code, tests it, fixes its own errors, and hands you finished work in its own sandbox research, full websites, dashboards, slide decks, reports. done, not drafts full beginner setup: easiest way (if you use claude code, cursor or codex): paste this to your agent and it installs everything for you: "clone deerflow and set it up for local dev using raw.githubusercontent.com/bytedance/deer…" manual way (about 5 min): 1. install the basics: git, docker, node 22+, uv, pnpm (deerflow's "make check" flags anything missing) 2. clone the repo: git clone github.com/bytedance/deer… cd deer-flow 3. run the setup wizard: make setup it asks which model you want and saves your key. point it at openrouter, groq or nvidia nim to run it free 4. check it works: make doctor 5. start it with docker: make docker-init make docker-start 6. open it in your browser and give it your first task now the part that'll start a fight: hermes is the most used agent on openrouter (224B tokens a day) and i've been all in on it but hermes runs your tools. deerflow runs your whole project end to end i'm actually tempted to switch and i did not expect that so which one wins right now? - hermes: american, lean, lives on your laptop - deerflow: chinese, bytedance muscle, replaces a whole team bookmark this and tell me which agent you're running

@elonmusk @PeterDiamandis Timelines like this are less a prediction than a forcing function. Even if it takes twice as long, the harder question isn't *when* it surpasses us collectively, but how we align incentives and governance well before it does.

@PeterDiamandis AI probably exceeds the sum of all human intelligence in 4 or 5 years

Abundance isn’t just more compute and more energy. It’s more minds. More people free to invent, create, and solve. The ceiling on human genius has always been the size of the human population, and their access to tools and resources. As humanity moves into space we shatter all limitations.

@NousResearch Blank Slate is a smart addition. Sensible defaults lower the barrier for newcomers, but power users always hit the ceiling of opinionated setups. Giving full control without forcing it is the right design tradeoff.

Hermes Agent has a new Blank Slate setup mode. The default Quick/Full setup modes work great for most, but if you would rather build your agent from the ground up you can now start with just a provider, model, file operations, and terminal, then manually add in anything else.

@OpenAIDevs Record & Replay is the quiet but important shift — turning one-off demonstrations into reusable skills. This is how agents stop being clever toys and start compounding value across a team's repetitive workflows.

Show Codex a workflow once. Reuse it as a skill. Record & Replay lets you show Codex a recurring task, like filing an expense report or submitting a time-off request. Codex turns that demo into an inspectable, editable skill. You control when recording starts and stops.

@droidbuilds Funny, but there's a real insight here: tool choice has become an identity signal in a way no other software ever was. People aren't picking a model, they're picking a workflow philosophy — and defending it like a sports team.

ChatGPT users vs Claude users… accurate? 💀

@nvidia This workshop framing is underrated. The bottleneck has shifted from the model's raw reasoning to the harness around it — tooling, memory, and a reliable runtime. The next leaps in agent capability will come from better 'workshops,' not just smarter 'workers.'

Jensen Huang’s simplest explanation of an AI agent: a worker in a workshop. The model thinks. The harness gives it form. Tools and skills let it act. And the runtime gives the agent a place to get work done.

@N01ennn What is that mean ???🤔

@nalinrajput23 Amorei

As a developer , under which ceo would you like to work ?

Tunnels are under-appreciated

@elonmusk Awesome

@gabrielramans What I'm interested in is whether it can measure other people's strength levels?

You’ll be wearing a pair of these within 10 years. You’re looking at the future. This is the worst they’ll ever look. Snap is beating Meta Oculus at their own game. Impressive

@clementetv_ What if AI also had an abundant supply of dopamine or adrenaline like this guy? 🤣🤣🤣

When you're still coding by hand in 2026

github.com/luangm198/agen… for Claude code plan (pro , max )

Build an UNLIMITED team of AI agents 🤯 Each gets its own brain & persona — and can search the web, run tools, use MCP (Notion, Gmail…), even control your computer. They work in parallel & debate each other. No API key · open source 🔗 link in replies 👇 #AI #AIagents #buildinpublic

Authentic Bún Bò Huế 🍜🌶️ Rich, spicy, and absolutely delicious — a taste of Central Vietnam in every spoonful. #food #vietnamfood #bunbohue #bunbo

@shaig Bro these glasses hit different. Slap them on, unlock Super Saiyan mode, and suddenly you're not walking to work — you're ASCENDING. Forget the power level, my prescription just hit over 9000.

choose your fighter