← Back to search
China's Step 5 DESTROYS Gemini 3.8 Flash?
AI News Today | Julian Goldie Podcast · 2026-10-11 · 6 min
Show full episode description
China's Step 5 DESTROYS Gemini 3.8 Flash?
✨ Episode Outline — click any point to jump to it in the episode
Problem solved
Finding a free, fast model for sub-agents and background automation in
Hermes Agent without paying for frontier models like
Claude Opus.
Benefits
- Free for one week across OpenCode, Cline, Nous Research, Kilo Code
- Plugs directly into Hermes Agent and the free AI coder as a model option
- Fast sub-agent model for background building and automation
- About 42% cheaper per task than Gemini 3.8 Flash while scoring higher
- Weights to be released open source on Hugging Face
Use cases
- Plugged Step 5 Preview into Hermes Agent as a free model for sub-agents and quick answers
- Added Step 5 to the free AI coder alongside 9Router, OmniRoot and previously Logs Alpha
- Run Step 5 side by side with Hermes Apollo, Hermes Oracle and Hermes Music workflows
- Use Hermes Cloud on a VPS so background automation doesn't touch the local setup
KPIs / results
- 600B total parameters, 27B active (mixture of experts), 1M token context, vision
- Hermes Index score 33.89, similar to GP6 Luna and HY4
- $0.72 per Artificial Analysis intelligence index task vs $1.24 for Gemini 3.8 Flash
- 100 output tokens/sec vs 331 for Gemini 3.8 Flash
Tools / build
- Hermes Agent with Step 5 Preview
- Agentic operating system (AI Profit Boardroom)
- Free AI coder with pluggable models
- Hermes Cloud on a VPS
- Hermes Apollo / Hermes Oracle / Hermes Music
We have a brand new model for free, which is Step 5 Preview. This is a brand new Chinese AI. And you can also use this for free for one week across OpenCode, Klein, News Research, Kilo Code. We've already plugged it into our Agentic operating system. So you can see that we can use Hermes Agent with Step 5 for free. Over here is a free model, which is pretty cool. And then we can build with it. And then we also have this inside our free AI coder. So we've been testing out as a free model inside here. And I'll talk you through what it is and how it works and why it's different, etc. Now, one thing to note here is it is not the most frontier model. So this is not going to replace Claude Opus or anything like that, right? It's not really designed for that. But what it is pretty good for is a free model for sub agents. Also, super fast. And you can use it as a free model for building and automating whenever you want in the background, right? So you can see, for example, here, we have the free AI coder that we've built in. And then we can plug any new model that comes out. So previously, we had Logs Alpha built into it. And we can use 9Router, OmniRoot, etc. with this. And then we can go inside the chat here and ask it to build something. Or we can go back to Hermes Agent. And over here, we can actually use it as a model. Now, this is pretty cool because we need really quick answers. Number one, Hermes Cloud is really good. But if we can use free models like StepFun, that's really good as well. Bear in mind as well, Hermes Cloud is basically on a VPS. So when you're using it, that means that it's set up on a VPS. It's not affecting your local setup. But you can plug it into an agentic operating system like this. And you can use it side by side with something like Hermes Apollo, as you can see right here that we've got ready to go. And also Hermes Oracle and Hermes Music, etc. You could use it across all of these different workflows, which is pretty nice. So let's talk about what it is, how it works, why it's different, etc. So number one, this is 600 billion total parameters with 27 billion active. So it's a mixture of experts model, but also is 1 million token context window. And it has vision as well. So you can see things when using it. Now, it's performed on the benchmarks at a similar level to GP6 Luna on Hermes Index. So if you're not familiar with Hermes Index, this is a new index where they test all the models recently. And it's scored 33.89. So this is not like GP6 Astra level, just to be clear here. But it is pretty good as a sub-agent model that works nice and fast, basically. And you can see how it performs on the leaderboards over here. So step 5 preview. It's around the same sort of level as HY4 and GP6 Luna. And it was built by StepFun AI. Also, something interesting to know here is they seem to have skipped step 4 and gone straight from step 3.7 to step 5. And it can work with text and images, 1 million token context window, mainly text output if you're using this. You have a performance on the benchmarks here as well. Now, if you're wondering, okay, where does this perform on the artificial analysis index, which is pretty useful for comparing the models? You can see it's right in the middle, right? So it's a similar model to Luna or Gemini 3.8 Flash, which means it's fast, easy to use, pretty cheap even if you don't have it on the free model. And also, it's outperforming a lot of models. So you can see, for example, Mistral Medium 3.5 over here. We use Glimmer, GP5.6 Luna. These are older models, but at the same time, not that old. Like, they just dropped the last couple of months, really. So it's beating Gemini 3.8 Flash on both intelligence and cost, which is pretty crazy when you look at a Chinese AI lab, like one that most people have never heard of. And it's pretty much right up there with Gemini 3.8 Flash and outperforming it, which is wild. That's like Google's best, fastest model right now until Argon comes out. Now, also, one of the big differences versus Google, and this is a great article, by the way, by OfficeChart. You can see here, it costs $0.72 per intelligence index task. If you're using it on the normal API, bear in mind, it's free for one week's test against $1.24 for Google's model. So it's about 42% less expensive to use something like this versus Gemini 3.8 Flash. But at the same time, it's outperforming it. So if you want to indicate how does it perform on intelligence, you can see, if we have a look at artificial intelligence analysis here, you've got step five preview over here. And then it's outperforming Kimi K3 Max and Gemini 3.8, and also DeepCity v4.1. It's just behind Grok 4.6. Obviously, Grok 4.7 is out now, but you get the point. Like, it's performing pretty well on the tasks versus other models. However, one thing to note here is that Gemini 3.8 Flash is actually faster. So Gemini 3.8 Flash is outputting at 331 output tokens per second, whereas Step 5 Preview is 100. So Step 5's model is still quicker than like AstraZix or Fable 5 or Opus 5. But bear in mind, it is not as fast as Gemini 3.8 Flash. Now, one thing to note as well is it's very verbose. So if you're using it, it's a cheaper token, but it also uses a lot of tokens. So if we have a look, for example, here, it produces a lot more than a typical model when it comes to output. Also, it is going to be open source. So they're planning to release Step 5 open source on Hugging Face, and it'll be open source for people to use. Obviously, it's a pretty big model. So most people are not going to run it locally, but it will be open source, which means people just get the weights openly, essentially. And you can get it available on OpenRouter, on News Research, and everything else. So thanks so much for watching. If you want to get more training on this sort of stuff, feel free to get that inside the AI Profit Boardroom. Link in the comment description or go to the AI Profit Boardroom.com. That is Step 5 Preview. And it's a decent model. It's not going to replace any Frontier models, but it's okay. It's fast, and you can access it for free for a week, which is pretty cool as well. Inside our community, you can post questions, get help and support. I personally create a lot of video tutorials for people inside here as well, and I just try and help you as much as you can. And then also inside the classroom, you can get access to all of our best training. So we drop new daily updates, as you can see right here, just to help you as much as we can. And we're always testing new things. We've also got our full agentic operating system where you can use free models. For example, as soon as Step 5 came out, we've already plugged it into Hermes, as you can see right here. And then you can also use it inside our free AI Coder, and it will plug into the memory system and everything else inside the Agent OS system. We literally just updated it today, so you get the zip file to install it and everything else. And then in the calendar, you can jump a weekly coaching course, ask questions, get help and support in real time. In the map, you can meet people in your local area. This is all available inside the AR Profit Boardroom. Link in the comments description, or just go to the ARProfitBallroom.com. Nice watching.