Previously, we covered the self-developed [product] launched by major phone manufacturer Xiaomi Mimo V2 Series Large Language ModelsWith over 1 trillion total parameters, it uses a mixture-of-experts architecture with 42 billion active parameters and supports context lengths up to 1 million tokens. It can easily handle complex tasks like large-scale codebase analysis and long document processing, and it’s also well-suited for agent-style work. Its token pricing is quite reasonable, but this model is practically inaccessible for non-China users, as purchasing requires real-name verification with a Chinese ID card and a Chinese bank account.
A free trial opportunity is here! From now until 4/22, simply go to Nous Research, link your credit card, and enable auto-reload to start using it. Nous Research is the company behind the recently trending Hermes Agent. Once you’ve set it up, you can use it with Hermes Agent.
We have partnered with @Xiaomi to bring their excellent MiMo V2 Pro model to Hermes Agent via the Nous Portal – completely free to use for the next 2 weeks!
Access now on the latest version of Hermes Agent: ‘hermes update’ pic.twitter.com/Fx7qQpvmOD
— Nous Research (@NousResearch) April 7, 2026
First, go to Nous Research pageChoose an account to sign in with, I’ll go with Google:
Next, select the Settings API:
Because this activity requires binding a credit card and enabling auto top-up, configure the settings in the Auto Top-Up section below:
Set up a payment method (Add Payment Method), and enter your credit card information to save it (don’t use it if you’re worried):
Next, enable auto-reload and set a minimum amount (5 USD) and a monthly maximum limit. I also set it to the minimum (10 USD), then save the settings:
The setup is complete here. If you want other Agents to use the Xiaomi Mimo V2 Pro, you can use “Create Key” below and give the generated API Key to an Agent like Openclaw, telling it to follow the instructions.Official Website Tutorial (Click here, paste the URL to the Agent)Just set it up (it’s free anyway even if you’re not sure):
Next, go back to your computer and type “hermes setup” in the terminal (make sure you’ve updated to the latest version of Hermes before setting up). This will take you to the setup interface (you can also do this during the initial installation, but beginners might not understand it). Here, select Model & Provider to set up a new LLM model, then choose Nous Portal:
Then, it should automatically open the Agent settings page on Nous Portal. Select the Free $0 plan:
Continue below, confirm that the Free plan is selected in the “SUBSCRIPTION” section, then click “SUBSCRIPTION AND CONNECT”. A confirmation window for the subscription plan will appear along with an SMS verification request. Normally, you will receive an SMS notification, and you can enter the verification code from it: 
The connection is established. You can close this page now.
At this point, your terminal screen will show model options. Nous offers many models to choose from, but they all cost money (pay-per-use, with a Coding Plan available as well, and you can use Nous Research’s own models, though I haven’t explored that yet). You can see that the free plan includes Xaiomi/mimo-V2-Pro available for use. After pressing Enter, Hermes Agent will automatically load and start using it:
At the end, it will ask if you want to set a Fallback model (a backup when your primary model fails). You can set it to your original model, or if you want to save money, you can set it to Openrouter Free. If you don’t know how to set it up, just press N. Remember to switch back to another model when it expires, or don’t come to me if it suddenly stops working one day:
Go back to Hermes Agent and enter /model to confirm if it’s using Xiaomi’s Mimo V2 Pro. Since it’s a $0 plan, you don’t need to worry about being charged:
After the free promotion ends, Mimo V2 Pro’s pricing is $1 per million tokens for input and $3 per million tokens for output, which is much cheaper compared to Claude Opus 4.6’s 5/25 (though it’s still more expensive than Minimax M2.7). Anyone interested can test its capabilities first before deciding whether to subscribe. My current tests show it’s quite good, but I’m still a bit hesitant since there’s no token-based plan available.
Source: KOCPC Chinese