Franck Mercado

Deepseek V4 Flash is now free via Nous Portal for a limited time thanks to @novita_labs!

Give our early preview of Computer Use (with ANY model) a try today! Built into the latest Hermes Agent and powered by @trycua - opens the door to any model, not just the frontier models in special modes - to control your actual computer. Best part, it doesnt take over your PC - you can continue to work and operate with full control of your keyboard, mouse, and screen - works entirely in the background!

anyone interested in or getting started with local ai personal inference, pay attention. start with the right practice. compile llama.cpp from source. i know lm studio and ollama exist. they're great onramps. but they're mostly wrappers around llama.cpp with abstraction layers that hide the flags you actually need to tune. what compiling once gets you: > the best inference engine for personal use, full stop > latest features the day they merge (vulkan flash attention dp4a, kv cache quant, fa toggles) > exact gpu arch optimization (sm_120 for 5090, sm_89 for 4090, sm_86 for 3090) > direct flag control > openai-compatible llama-server api ready out of the box the build (3-5 minutes on a modern cpu): git clone github.com/ggerganov/llam… cd llama.cpp cmake -B build -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=120 cmake --build build --config Release -j (replace 120 with 86 for 3090, 89 for 4090, 80 for A100. for AMD GPUs swap GGML_CUDA for GGML_VULKAN.) when to NOT use llama.cpp: > multi-gpu batch serving at scale = vllm > production async high-throughput = vllm or sglang > apple silicon = mlx is faster for single-gpu personal inference + agentic workflows + benchmarking: llama.cpp from source. every time.

@diego_defi @opencode flash y pro ya está.

Hace unos días empecé a usar @opencode Go con Hermes, depende de la tarea es el modelo que uso.

Local AI is having its moment! Below is the number of new GGUF models created each month over the past 8 months & insights from our HF internal agent (May is partial): - 176,000 total public GGUF models on HF - Two distinct regimes: Oct–Feb averaged ~5.1K new GGUF models/month. Then March–April jumped to ~9.2K/month — nearly double the previous rate. - March was the inflection point (+55% MoM) — likely driven by a wave of new open-weight model releases being quantized to GGUF. - April sustained the momentum at 9.7K, suggesting this isn't a one-off spike but a new baseline. - The GGUF ecosystem is accelerating — the community is quantizing models faster than ever, likely thanks to better tooling (llama.cpp improvements, automated quantization pipelines, and more models supporting GGUF natively). Let's go!

Hermes Agent is now #1 on the Global @OpenRouter token rankings. While our journey together has just begun, we'd like to take this opportunity to thank our contributors, supporters, and users for all they have done to get us this far.

@fashionluvr @VadimStrizheus what models would you recommend?

@VadimStrizheus It’s nearly free on open router Why need to run local ?

To all of the local model pros out there, whats the most efficient way to run Gemma 4? I have a: 16Gb 512 M4 Mac Mini 24GB 1TB M5 MacBook Pro I’m running Hermes agent on my Mac Mini with Claude and Codex rn, and nothing on my MacBook Pro rn. also what the heck is turbo quant? Thanks yall

holy shit. within a month nous research went from 4.5k gitHub stars to 100k and completely ate the lobster. now climbing past 100k and going to eat the rest as the top general agent from science to content to philosophy to code, all baked into one hermes agent. the messenger of intellect (nous) pervades everywhere.

@Scobleizer Could we get audio or transcription of this X space?

NTS: Opus 4.6 is considered to be a form of "minimum viable AGI," demonstrating capabilities like novel reasoning and goal achievement,but it still lacks full human-like generalization and flexibility. Current benchmarks show that it struggles w tasks requiring dynamic learning.

This is nuts: Clawdbot figured out how to transcribe and respond to a voice message on its own, detecting the Opus format, converting it via FFmpeg, calling OpenAI’s Whisper with a found API key, and replying as if voice support had always existed.

So now, with AI, the industry will be shifting away more rapidly from the traditional "request-response" cycle to binaries-first alternatives 🙀 And the clear winners are Go, Rust, PHP, JS, Java and leaving behind Ruby, Python (maybe not with Mojo), Perl, Vanilla C++

@jessegenet @openclaw an agent per kid like a companion bot that adapts to his/her progress?

Pulse check. How much effort should I put into sharing how I’m using @openclaw to homeschool my kids? Is this content anyone wants!?

@JJeffrey100 @jessegenet @openclaw which startups?

@jessegenet @openclaw Further, have openclaw build a platform for parents to learn, share, and collaboratively improve curriculum. There are a couple startups doing this, but they're charging $10k/yr. You could build it for $200 subscription to max. Or, I can help and use my max subs

@aravindballa @SimonHoiberg agree but n8n for non-techs maybe?

@SimonHoiberg tbh, you don't need n8n when you have openclaw. openclaw can even do cron jobs, respond to webhooks and everything.

OpenClaw + self-hosted n8n is so powerful my mind almost cannot comprehend it. And there are still people who don't see what the hype is all about... I'm sorry... I can't help you 😅 You either get it, or... good luck to you in the upcoming few years.

the whole reason to self host IS TO USE A LOCAL LLM so your API keys, passwords, emails, calendar, health records, business data, and everything else are not sent to an API provider like OpenAI, OpenRouter, or Anthropic Mac minis are NOT GOOD for that, BUT A GPU IS Buy a GPU

When kids leave traditional school, they have to go through a deschooling process. Deschooling = unlearning all the bad habits you picked up in school. Some of the biggest habits kids need to unlearn:

We spun up a new GitHub repo for all things MCP at @Google. Get info on our remote managed MCP servers, open source MCP servers, examples, and learning resources. github.com/google/mcp

This is sick 🤯🤯🤯 You can monitor Claude Code tasks using a project management app Install the @linear MCP then ask Claude Code to make a plan first then save tasks to Linear. Ask it to update once each task is finished $ claude mcp add --transport sse linear-server mcp.linear.app/sse

All good with AI until they hire an AI micromanager 💀