← Back to search
Hermes Agent + Step 5 is INSANE (FREE!)
AI News Today | Julian Goldie Podcast · 2026-10-11 · 7 min
Show full episode description
Hermes Agent + Step 5 is INSANE (FREE!)
✨ Episode Outline — click any point to jump to it in the episode
Problem solved
Using the free Step 5 Preview model inside
Hermes Agent for fast agentic tasks, skill learning and multi-workflow automation without a frontier model.
Benefits
- Step 5 Preview free for one week via Nous Portal, open weights coming
- Switch models inside the agentic OS and chat with Hermes directly
- Teach Hermes new skills from a guide with /learn
- Built-in web search via Perplexity's model in Hermes Cloud, no Firecrawl API needed
- 1M token context holds lots of memory in one agentic session
Use cases
- Used /learn on a tutorial guide so Hermes drafted an 'AgentOS dashboard' skill.md capturing the system architecture
- Asked Hermes with Step 5 to research the latest Hermes Agent updates for the week
- Hermes Apollo voice agent (Codex, on a ChatGPT subscription) brainstormed Hermes SEO ideas: rankings, competitors, keyword gaps, meta tags
- AgentOS SEO content written with Claude and Hermes ranking in Google AI views for 'best harrow agency'
- Obsidian 'memory galaxy' syncing memories from Apollo, Oracle and Radar sessions
KPIs / results
- 600B total parameters, 27B active, sparse mixture of experts
- 1M token context window
- Free for one week; on par with Gemini 3.8 Flash and cheaper
Hermes Agent just got a brand new free model, which is Step 5 Preview. So you can use this on Nuse Portal for free as a cloud-based model. It's also going to be OpenWeights and I've already plugged it into our Gentic operating system like you see right here. So we can switch between the different models. We can choose between them as you can see right here and we can choose Step 5. Then we can say something like, I don't know, plan my day or research topic, draft something or something like that. We can plug in, hit send and start using Hermes with Step 5 as you can see. Now the awesome thing about this is that it's free to use. You can plug it into Hermes. It's super fast to use, so you can see right here. And it's quick and easy as well. So let's say for example, we wanted to teach Hermes about our Gentic operating system. What we could do here is we could get this guide, as you can see right here. Then we can go back to the HNOS. We could type in forward slash learn, plug in the tutorial guide, hit send. And then we can actually teach Hermes agent a skill from the guide using this system right here. Now you can see we've taught it the new skill. So skill name, AgentOS dashboard. It's drafted the skill MD file here and then it's captured the whole core architecture of the AgentOS and we've taught it the whole system right here. Pretty cool. And then if we're like, okay, I want you to build an AgentOS or I want you to improve mine, et cetera. We could easily go with this directly. The other cool thing is here, like we could actually chat to it inside the AgentOS. So for example, I could say, okay, can you research the latest Hermes agent updates this week and how I could use them? So we can ask you, but then also what we can do here is we can open up Hermes Apollo, which is a voice agent plugged into Codex. So we can start speaking to it as you can see right here. Good afternoon, sir. How may I assist? Could you give me some ideas on what we could do with Hermes agent for SEO? Certainly, sir. You might have Hermes monitor rankings and competitors, surface keyword gaps and propose content updates from search console data. Okay, perfect. It could even draft a meta tax. Very good, sir. Just going to close out right now, but you can see how you can basically use Hermes Apollo whilst you're building with Hermes agent, which means if you have any quick questions in between, we can actually use a voice agent from Codex here. And that doesn't cost us anything because we're using our chat GPT subscription plugged into this. Whilst we have step fun plugged in the middle here, right? And it's just super fun to use Hermes agent this way. Now you could use it inside the desktop, but I just don't find the UI fits exactly what I want. Whereas this system right here, it all plugs into one place and we've got Claude and we've got anti-gravity. We've got our memory system all plugged in one place. So yeah, it's pretty good as API here. I think it's very good for agentic stuff because normally you don't need a frontier model for agentic tasks unless you're coding out something super crazy. But then normally you would just use like Claude CLI or Codex directly instead. And with this system, it's really quick to reply, right? So one of the best things about step 3.5 flash is, sorry, step 5 preview is that it is lightweight, designed to be fast. It's not frontier level, but it's right on par with something like Gemini 3.8 flash and it's cheaper and you can use it for free right now for one week as well. And you see that actually does pretty good research here. So it's pulled in for example, okay, here's a new update on Hermes agent. Here's some of the new things that have been plugged in. Here's how we could switch and choose between them. And here's how we could actually use this in the future. And you can actually see that picked up the step 5 preview has just dropped inside Hermes agent here as well. It's really cool. And then we could use it across multiple workflows here. So not only does it have web search plugged in directly into Hermes cloud, because they're using perplexity's new model for web search, which means you don't need to plug in a Firecrawl API or anything like that. But also we can switch over and we could start using Hermes Apollo over here. We could use Hermes Oracle, which basically scans the latest news and gives us all the latest information on what's happening inside our industry. Over here, we could switch over to Hermes Astros, which is monitoring our competitors. And we could use Hermes Oracle as well to just find out what's trending in our industry. And the whole point of this is you can combine multiple workflows. So for example, what you could also do is you could go over to the SEO agent here and you could use a model like step 5 preview. No problem because it's fast. It has web search enabled with Hermes agent. It's free to use. And also you don't need a frontier model for writing content, right? You just need to give it the right workflows here. And that's basically what we use in per AI SEO. In fact, if we type, for example, something like best harrow agency, you can see that our content from our agent OS written with Claude and Hermes is ranking really nicely inside Google AI views here. So you can see us ranking over here and then also here. And so we have a good system. You could plug this into your SEO workflows. You could use this with the video agent. You could use this directly inside the Hermes chat. However you want to use it, it's ready to go. Really powerful stuff. Now, if you're wondering, okay, where is step 5 preview from? So it comes from an AI coding lab in China called Step Fun. And it's Step Fun's flagship foundation model. It's 600 billion parameters total, 27 billion active parameters. So it's like a sparse mixture of experts model. And it's got a million token context window. So it's pretty good for a genetic task because it can hold a lot of context and memory in one session. And also it can directly work with your memory systems, right? So for example, if we go to Obsidian over here, we've got our memory system, right? This is our memory galaxy. And it's got all of our memories stored in one place, which we can easily access. But if we go to, for example, our recent memories here, you can see that it's pulling in the recent memories when we were talking to Hermes Apollo or when we were talking to the AI news researcher, the Oracle, the radar system. And so it's all plugged into one and it all syncs up really nicely. So if you want to get this full system, you can get that inside the AI profit boardroom. This is an agentic operating system that I've built, the worst with Hermes agent. And also inside the AI profit boardroom, you can get access to all of our best trainings, tutorials, tips. We actually have a lot of training on Hermes. So if you type in Hermes in the search bar here, you can see all of our new blueprints and automations with Hermes agent. And also inside the community, you can post questions, get help and support in real time. And I personally answer them often with video tutorials and everything else. And then in the calendar, you can jump a weekly coaching course, ask questions, get help and support in real time. Inside the classroom here, if you go to the new daily updates section, we have the full agentic operating system with a zip file that you can give to your agents to install. And then we also have new trainings on Hermes and everything else that we test daily as well, just to help you as much as we can. Now, additionally, if you're a complete beginner, you can get our beginner to expert roadmap here and we show you exactly what to do week by week. And then also, if you want to get more training on Hermes, you can get over here. So we have a full Hermes agent course that you can see. Plus, we have four weekly coaching calls and a map where you can meet people locally near you, put building out with AI agents like you. So feel free to get this link in the comment description or go to the AIprofitborn.com. Nice watching.