China’s Minimax M3 + Hermes AI Agent is INSANE!
The video demonstrates how to combine China's newly released Minimax M3 AI model with the Hermes Agent to create a free, autonomous AI worker capable of running tasks for up to 12 hours. The presenter walks through installation, live testing of agentic tasks like opening apps and scheduling, and explains the brain-body analogy between Minimax M3 and Hermes. A cloud-hosted alternative called Max Hermes is also introduced.
Summary
The video introduces Minimax M3, a newly released coding and agentic AI model from China, and demonstrates how to integrate it with Hermes Agent to create a powerful, free AI automation system. The presenter begins by outlining the core value proposition: a free AI worker that can run autonomously for 12 hours, handling tasks like web browsing, app control, scheduling, and report sending.
The setup process is explained step-by-step, requiring Hermes Agent and Ollama to be installed and running. A simple terminal command — 'ollama launch Hermes' with Minimax M3 specified as the model — connects the two systems via Minimax M3's cloud. The presenter notes that the free Ollama plan has token limits but most users won't hit them, and that one-click integration with tools like Claude Code, Codex, and OpenClaw is also available.
Live demonstrations show Hermes performing agentic tasks: navigating to websites, opening the Notes app and typing a message autonomously, and scheduling recurring calendar events. The presenter acknowledges that performance through Ollama can be slow due to cloud routing, and recommends the Minimax coding plan for faster performance.
The brain-body analogy is used to explain the system: Minimax M3 is the 'thinking engine' or brain — a self-learning, self-improving model designed for long-horizon agentic tasks — while Hermes Agent, created by Nous Research, serves as the 'body' with an autonomous self-learning loop that improves with use. Together, they form a free, open-source agentic system capable of content writing, web search, email sending, app control, scheduled task execution, and business learning over time.
Benchmark comparisons show Minimax M3 is competitive but not quite at the level of Claude Opus 4.7 or GPT 5.5, though it is significantly cheaper and purpose-built for agentic use. It also supports MCP tools, has a 500,000 token context window, accepts text and image input, and is described as an open-weights model with plans to be fully open-sourced on Hugging Face.
A cloud-hosted version called Max Hermes is introduced as an alternative for users who want to deploy without a local setup, though the presenter personally prefers his custom Agent OS system for its superior UI and live preview capabilities. The video closes with a pitch for the presenter's paid community, the AI Profit Board, which includes guides, prompts, coaching calls, and his full Agent OS setup.
Key Insights
- The presenter claims Minimax M3 is a self-learning, self-improving model designed specifically for agentic capabilities, capable of running autonomously for 12 hours on long-horizon tasks.
- The presenter states that Hermes Agent, created by Nous Research, has an autonomous self-learning loop, meaning the more it is used, the better it gets — a key reason it pairs well with Minimax M3.
- The presenter demonstrates that Hermes can perform agentic tasks — such as opening the Notes app and typing a message — entirely in the background while the user continues interacting with it in the foreground.
- The presenter notes that Minimax M3 is not at the benchmark level of Claude Opus 4.7 or GPT 5.5 on SWE benchmarks, but argues it is significantly cheaper and purpose-built for agentic tasks, making the trade-off worthwhile.
- The presenter reveals that Minimax M3 is an open-weights model and states that a fully open-sourced version on Hugging Face is coming soon, distinguishing it from closed frontier models.
Topics
Transcript
[0:00] Miniax M3 just dropped out of China. When you plug it into Hermes Agent, you get a free AI worker that can run on your computer for 12 hours straight. We're talking about an AI that can open up your app, schedule task, search the web, send reports all on its own using this powerful agentic model that literally just dropped. And in this video, I'm going to show you how to set up, test it live, and real task. And there's one thing you absolutely have to do to unlock the full power of this. Plus, I'll show you something really [0:30] cool at the end, which is a cloud hosted version of this you can access from…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Julian Goldie SEO
Google AI Studio Just Got A Huge Security Upgrade
Google is reportedly developing a new "Security Check" mode for AI Studio that would apply secure coding practices to AI-generated applications. The feature aims to automate security verification during app development, addressing common vulnerabilities like exposed keys, weak authentication, and data access issues before deployment.
New Gemini Skills Update Is WILD!
Google has launched Gemini Skills, a new feature allowing users to save custom instructions once and reuse them repeatedly by typing a slash command, replacing the discontinued Gems feature. Skills can include reference files, be combined together, and are built on an open file format (skill.md) that allows portability across different AI tools.
NEW Codex CLI Update Is ABSURD!
OpenAI's new Codex CLI update transforms terminal-based programming assistance by enabling multiple AI agents to work simultaneously on different tasks, visible in a full-screen control center interface. The update includes features like Agent View, work trees, voice control, and parallel task execution, fundamentally changing how AI assistance operates from sequential Q&A to autonomous team-based work management.
NEW Hermes Cloud + Agent OS is Crazy! 🤯
Hermes Cloud combined with Agent OS enables users to deploy AI agents in the cloud 24/7 using free models without API keys or local setup complexity. The system provides persistent memory, sandbox isolation, and integration with custom workflows like content research, competitor analysis, and news monitoring.
Manus 2.0 Just Changed AI Agents Forever
Manas 2.0 represents a fundamental architectural overhaul that enables users to create games, websites, videos, and automations through simple text requests, with the key innovation being editable components rather than fixed outputs. The platform uses a new Cascade engine that reduces token usage by 23% and improves task speed by 28%, while maintaining all project elements in one connected workspace.