Eric Henderson

@grok and @suno helped us make this music video in under 1 day. BLANKET MEN (Cape of Concrete) youtu.be/chkHki32PSI?si… via @YouTube

@cecegkh Miranda?

Police officer approaches the woman and starts asking questions about what’s in her stroller. She starts off soft and kind of shy, but you can literally hear the moment she knew she was caught. Everything about her changed. She refused to listen to him because she didn’t like how he was talking to her. Then she started the usual “Don’t touch me” bullshit. Why do people think officers don’t have the right to touch them? If it’s a lawful arrest, the police have every right to cuff you.

@SaltyBitch_52 Just report to police and door dash after you have her on camera admit she for the food. Don't make a scene even though it's infuriating...

This DoorDasher got a notification that the order he delivered to the hospital was never received by the customer so he went back to confront the lady. He accused her of trying to get free food when he gave her the bag of items in front of other people. By her reporting that she didn't receive the items, the driver gets flagged as not a good fit for the company. After several reports, you lose your job. The lady eventually admits she got the food and the driver wants her to make it right because he's trying to make extra money to feed his family. Do you think that he did the right thing by going back to confront the customer?

more specs

🚨 BREAKING: CHINA just released a Python framework for building AI agents. 100% OPEN SOURCE. It has visual agent design, MCP tools, memory, RAG, and reasoning. All built in. All working together. It's called AgentScope. You describe your agent system. It builds the architecture, wires the tools, and runs the whole thing. You come back and there's a working multi-agent pipeline. Not a prototype. Not a demo. The actual system. Not a wrapper. Not a chatbot builder. A full Agent-Oriented Programming framework that thinks in agents from the ground up. Here's what it does out of the box: → Visual agent builder so you design your entire system before writing a single line of code → Native MCP tool support, plug any external tool directly into any agent in your pipeline → Built-in memory so every agent remembers context, decisions, and history across sessions → RAG pipeline ready to connect your own documents, databases, and knowledge bases → Reasoning modules that let agents plan, reflect, and self-correct without human input → Multi-agent coordination so your agents collaborate as a system, not a pile of isolated API calls Here's how it thinks: You define your goal. AgentScope maps the agent roles. Each agent gets its tools, its memory, its reasoning layer. They coordinate. Results flow back up. You get a finished output. A single complex task might route through a planner agent, a researcher agent, a coder agent, and a critic agent, each doing its job, then converge into one clean deliverable. Here's the wildest part: AgentScope is built by Alibaba DAMO Academy. The same lab behind Qwen. They didn't assemble this from existing pieces. They designed the entire framework from first principles around how agents actually need to think, remember, and work together. Most frameworks give you building blocks. AgentScope gives you an architecture. The community has already started plugging it into data pipelines, research workflows, and full automation systems the team never planned for. 100% Open Source. Apache 2.0 License.

@elonmusk But why take up so much space? Build another 4 levels up instead of 400x wider and longer...

That will be one hell of a building!

Thank you for building it.. if you build it we will buy it, right? We will have the option to buy at least... When we don't have any options, out country has lost... If we lose then humanity loses... But what's going to happen to help my generation and my kids earn enough now while we still can before we're all replaced? I'm laying down new code and learning new skills daily but I won't be able to forever ...

@MaryamF12389 I have. I married my Highschool girlfriend and are still here 28 years later. I'm 100% neither of us have strayed ... Sadly 60% or more of marriages fail now. People live and do dumb things and that's humanity but it doesn't mean we're incapable of choosing to not cheat.

Be honest....???

@Alexintosh Just trying to follow along made an explainer for me. Maybe skip the last minute.

I just ran Qwen3.5 35B on my iPhone at 5.6 tok/sec. Fully on-device. 4bit | 256 experts. Model: 19.5GB. iPhone: 12GB RAM. wild.

Look at these and read them carefully, this will give you an idea of what you can expect from different Open Weight models. Qwen3.5-27B is the smartest Qwen3.5-35B-3A is the fastest huggingface.co/Qwen/Qwen3.5-2… huggingface.co/Qwen/Qwen3.5-3… Any Laptop, Macbook, or GPU should handle em

@javierblez @NousResearch Went to log in to your app with X connector but not willing to give you access to all private posts too .. come on

@NousResearch Try the game here: warv3finale.com - I’d love any feedback or suggestions!

This broke my mental model of game dev 💀 2.5 hours → fully playable ‘Worms’ clone. Built with Hermes agent by @NousResearch Here’s what made that speed possible: Hermes used ‘Persistent Shell’ mode, which ensured it didn't forget its current folder or active tools. This allowed it to work smoothly, without the distraction of constantly having to recall where it left off last time. To optimize the workflow, the agent moved beyond linear execution and parallelized the workload. It spawned isolated subagents while executing multiple independent tool calls via ThreadPoolExecutor. Like, one subagent wrote Python RPC scripts for the projectile physics while another utilized vision tools for character sprites. When the complex terrain logic required debugging, the agent used filesystem checkpoints and the /rollback command to instantly return to a stable state. To fix UI bugs, it attached to a live Chrome instance via CDP (/browser connect), fixing rendering issues in real-time. The agent’s built-in learning loop was active from the very beginning. By the time the game was finished, this continuous process allowed the agent to autonomously convert the physics logic into a custom skill. This logic is now a permanent plugin file in the agent's plugin architecture, making the physics engine a native capability that the agent can reuse for future projects. Follow @warv3finale for updates!

@javierblez @NousResearch Very cool. Will you share? I'm comparing openclaw to Hermes this week...

Qwen 3.5 27b, llama.cpp server, hermes agent, tailscale, 3090 The stack remains undefeated Testing all inference engines, after that seeing if there are any model specific tweaks we can make to get it running faster After that maybe maybe mess around with serving w/ tinygrad just for fun

Hey Peter, I need a set of h100 or at least some 5090's from that shop in China that puts more GB on each card... Because Chase said I need to add your obviously responsible for making a product that I need better stuff to use more efficiently... Can I have my time back from all the help I have given in discord to your customers? OMG the nerve of people these days ..

This guy emailed me asking for a *token session refund* because his claw made mistakes. 🙃

@stevibe Amazing

Part 2 by popular demand. Qwen3.5 2B vs 4B vs 9B on the same screenshot-to-code challenge. At what size does it break?

"122B has to be smarter than 27B" I showed 4 UI components to three Qwen3.5 models and asked them to recreate them from a screenshot alone: - 27B (dense) - 35B-A3B (MoE) - 122B-A10B (MoE) Same screenshot. Same prompt. Same task. Which one do you think nailed it?

You can now train Qwen3.5 with RL in our free notebook! You just need 8GB VRAM to RL Qwen3.5-2B locally! Qwen3.5 will learn to solve math problems autonomously via vision GRPO. RL Guide: unsloth.ai/docs/get-start… GitHub: github.com/unslothai/unsl… Qwen3-4B: colab.research.google.com/github/unsloth…

NVIDIA's Kimodo is the release of the week 🔥 Prompt the timeline whatever your want like: "a person walks forward" → "a person starts jumping", hit Generate, and watch a 3D character do it in seconds (700hrs of pro mocap training. Works on human + robot skeletons. Super fast + free to use on HF)