01What Does Run Hermes Free Forever Mean?

Run Hermes free forever means you can use the Hermes agent without paying for API costs or subscriptions by connecting it to a free local model or free API. In this episode, Julian Goldie shows how to do this with Google's Gemma 4 model, which is now fast enough for real agent work thanks to a new update.

The update works with Ollama on Apple Silicon using MLX, which makes Gemma 4 run 90% faster. This solves the old problem where local models were too slow to be useful. Now you can run Hermes in the background, have it build apps, learn skills, and run loops—all for free and with full privacy.

If you don't have Apple Silicon, you can still run Hermes free forever using a free API from OpenRouter, which offers the 31B model. This means the solution works for both Mac users and others.

02Key Features of Running Hermes with Gemma 4

  • Free forever: No API costs, no subscriptions, no token worries.
  • 90% faster with MLX on Apple Silicon.
  • Private and offline: Everything stays on your machine, works without internet (e.g., on a flight).
  • Agentic loops: Set a goal, Hermes loops until done, can check its own work and retry.
  • Sub-agents: Hermes can split tasks into multiple sub-agents working in parallel.
  • Skill learning: Use /learn to teach Hermes from tutorials, and it remembers forever.

These features make Hermes a powerful agent that can run 24/7 in the background, building stuff or watching news for you.

03How to Set Up Hermes Free with Gemma 4

  1. Install Ollama and make sure you have the latest version.
  2. Go to the models section in Ollama and find Gemma 4.
  3. Choose the MLX model (these are the new ones, not the old normal models).
  4. Run the terminal command: ollama run gemma4 (or the exact model name shown in Ollama).
  5. In Hermes, plug in the local model by selecting it from your profiles or adding it as an API endpoint.

If you don't have Apple Silicon, you can instead use a free API from OpenRouter that has the Gemma 4 31B model. Just plug that API key into Hermes, similar to other APIs.

04How to Use Hermes with Gemma 4: Examples

Julian shows several practical uses. First, you can ask Hermes to build a to-do list app locally, and it does it in a few minutes, showing tokens used and preview.

Second, use /learn command: give it a tutorial URL, Hermes reads it and adds that as a skill, so it never forgets. This works in the background on a free local model.

Third, you can point it to a file and ask it to read a landing page and list its sections. Fourth, you can have it check your latest emails and give you a lowdown. Fifth, set up scheduled loops to watch AI automation news and come back to you.

For loops, you set a goal, choose a builder model (like Gemma 4), and use a frontier model as a judge to decide if the work is done. You can set it to loop up to 20 times until done.

05Results and Comparisons from the Episode

Julian compares the old vs new: previously, Gemma 4 was too slow and unusable for agentic work; with MLX, it's 90% faster and works well. He shows examples of tasks completed: building a to-do app, learning a skill from a tutorial, critiquing and improving a tagline in a loop, and reading a landing page.

He also notes that even with frontier APIs, complex tasks like setting up a skill can take a few minutes; with Gemma 4, it's similar but you don't have to wait because it runs in the background.

Julian mentions he used to avoid Hermes for agentic tasks due to slowness, but now with this system, he runs it in the background without worry.

06Tips, Limits, and Mistakes to Avoid

  • Don't expect instant responses: It's an agent, not a chat bar; it works in the background.
  • Pick the MLX models, not the old normal ones, else you won't get the speed boost.
  • Loops save tokens: Old way required watching every step; new loops run automatically.
  • You don't need to be technical: Just two commands from Ollama.
  • Use sub-agents wisely: You can parallelize tasks for efficiency.
  • If you don't have Apple Silicon, use the free OpenRouter API instead of giving up.

Also, remember that anything you do locally stays private and offline, which is a major benefit.

07Who Should Use Hermes Free with Gemma 4

This is for anyone who wants to use AI agents without ongoing costs. It's great for individuals or small teams who want to experiment with agentic workflows, build apps, learn new skills, or automate tasks without a big budget.

Julian notes that even non-technical people can set it up; he has 94 pages of testimonials from people winning, learning, and growing with agent operating systems. He says if he can do it non-technically, anyone can.

Why it matters: It makes powerful agentic AI accessible to everyone, not just big companies with deep pockets. You can run Hermes 24/7, have it build stuff, watch news, or manage emails—all for free, privately, and without internet dependence.

08Get the Full Agent OS and Community

Julian mentions that inside the AI Profit Boardroom, you can get his full agent operating system, which includes the local loop engine, loop agents, playbooks to reduce token usage, and a memory system. He also has video tutorials for every Hermes update.

There are weekly coaching calls, a community for questions, and 24/7 support. You can access it via aiprofitboardroom.com or links in the description. If you want to learn more about Hermes, they create step-by-step guides for each new feature.

This episode's solution is a practical way to start using AI agents for free, and the community is there to help you succeed.