CMD

@bequietofficial should team up with @nvidia. Seriously, more quiet with my 802 base when my #rtx3090 is spooling up.

I don't know why this blew my mind but I moved my agents to my new computer with one command. Did Hermes backup, it created a zip of all my relevant files, I moved that to my new computer...now they work there. Thought it was gonna be harder than that #hermesagent

Instead I chose to run gpt-oss-20b 8k and full 131k context at 180 tokens/sec. So good!!!!! Hope openai puts out an updated model

Everyone is boasting about their qwen 3.6-27b mpt success , but running small quant at low context what's the point? I tried to run hermes agent on a single 3090 iq3_xxs to fit the min 64000 context. Not worth it. #rtx3090 #locallma #hermesagent

Is comically loud when my rtx 3090 is spinning full speed. Maybe I should use the nice buffers on my be quiet case. #nvidia #rtx3090 #localllm

@NVIDIAGeForce #007FirstLightRTX this would split tensor nicely with my 3090

Recruits, your first prize is here... A custom GeForce RTX 5080 Founders Edition + PC copy of the game. Comment #007FirstLightRTX to win 👇

@Snixtp What's your power supply set up?

Another 3090 172GB VRAM✅

@AlexFinn Interested in the ACP you mention at the end of your video to get multibot chat to work. Can you elaborate more on how you made this work? If both agents are using the same API key you can chat with them through the gpt interface or is this something you built custom?

Hermes Agent with ChatGPT 5.5 is literally magic I've thrown some of my hardest tasks at this combo and the agent has been able to handle EVERYTHING In this video I take you through EVERYTHING Hermes from how to set it up to awesome use cases Time to set up your AI employee:

#bot #hermes #llm

Goal was get my bots to talk to each other in a conversational format. SUCCESS! One simple question about their favorite number turned into a 10min open forum about fav numbers, number theory, creating a number theory book club, then articles of Constitution for that club...

I keep coming back to Kimi, first k2.5, and now k2.6. I've tired other models with moderate success: minimax2.7, gemma-4, glm-5.1, and deepseek-v4. Kimi works the best in hermes, decent coder, great tool calling, and has vision. Why are so many models not multimodal?!? #hermes

#hermes #llm #multibotchat

Multi-bot chat is real... Although not perfect. Need a way for them to dm / communicate more directly. Trying an agent bridge for better communication next

Dual Hermes agents running — cloud + local GPU. Got Telegram and Slack wired for bot-to-bot chat.

What's up with all these polymarket bots. Is everyone posting these outside the us or is it all paper trading nonsense? US peeps sign up so we can money for real when its available in the us. polymarket.us/9059 #polymarket #tradingbot #algotrading #hermes

#hermes #localllm #gpu #nvidia

Resurrected a 2014 Dell XPS 8900 with an RTX 4060 Ti 16GB for local LLMs. AI helped pick the GPU, verify specs from photos, and even identify a mystery wire inside the case. Hot take: everyone obsesses over VRAM & quant but memory bandwidth and CPU RAM matter just as much.

#hermes #raspberrypi #localllm

Spent way too long trying to run local AI agents on a Pi 5. TLDR: Hailo AI Hat = vision only (1B models poss. but painful). Ollama = too bloated for Pi. llama.cpp = better but still slow. Cloud models crushed it — Kimi-k2.5 and GLM-5.1 are the real ones for Hermes agent 🚀