Your Hermes agent
Running in minutes
VPS with Hermes installed and ready to configure. Choose your AI provider during setup. Always on. One sign-up. Full root access.
Your server. Your data. Cancel any time.
Introducing MiMo 2.5 Pro UltraSpeed
Watch MiMo UltraSpeed build a complete e-commerce product page with shopping cart, image gallery, reviews and related products in 60 seconds.
Running Hermes on a VPS is never just one thing
- 1Provision a server
- 2Install Hermes
- 3Open accounts with multiple AI providers
- 4Top up your balance
- 5Connect the service. Hope it works.
That's before your agent does anything.
Hermes.fast skips all of it.
One sign-up. Your agent is running
Pick your plan
Choose a VPS size based on your workload. Solo, Pro, or Power. Hermes is pre-installed and ready for your preferred AI provider.
Your server is ready
SSH in straight away with full root access. Hermes is already running. No install scripts, no config files. Tweak what you want, or leave it and start immediately.
Start using it
Connect Telegram, Discord, or your tools of choice. Your agent runs 24/7, even when your laptop is closed.
Simple pricing. Nothing hidden
You pay for two things: your server and the inference you use. That's it.
Your server
Solo
2 vCPU · 4GB RAM · 40GB SSD
Best for: Personal agent, 1 to 2 platforms
- Hermes pre-installed
- $5 of inference credit every month
- Always on, 24/7
- Usage dashboard
Pro
4 vCPU · 8GB RAM · 80GB SSD
Best for: Power users, multiple tools
- Hermes pre-installed
- $10 of inference credit every month
- Always on, 24/7
- Usage dashboard
- Priority provisioning
Power
8 vCPU · 16GB RAM · 160GB SSD
Best for: Heavy workloads, multiple agents
- Hermes pre-installed
- $20 of inference credit every month
- Always on, 24/7
- Usage dashboard
- Priority support
Every plan includes monthly inference credit. Top up any time.
Your inference
We pass DeepSeek, GLM, and MiMo cost straight through, plus a flat 10% to cover payment processing and currency conversion. Nothing else.
DeepSeek V4 Flash
Input / 1M
$0.154
Cached / 1M
$0.003
Output / 1M
$0.308
Everyday tasks. Conversations, memory, scheduling, Telegram, Discord.
DeepSeek V4 Pro
Input / 1M
$0.479
Cached / 1M
$0.004
Output / 1M
$0.957
Deep reasoning. Complex coding, research, long analytical tasks.
MiMo V2.5 Pro
Input / 1M
$0.508
Cached / 1M
$0.004
Output / 1M
$1.015
Xiaomi's flagship agentic model. Strong coding and long-horizon tasks at a low price.
MiMo V2.5 Pro UltraSpeed
Input / 1M
$1.523
Cached / 1M
$0.013
Output / 1M
$3.046
The same MiMo Pro served at up to 1,000 tokens/second, for agents that can't wait.
GLM 5.2
Input / 1M
$1.54
Cached / 1M
$0.154
Output / 1M
$4.84
Advanced multi-step reasoning, deep research, and complex analytical workloads.
GLM 5.2 Fast
Input / 1M
$2.31
Cached / 1M
$0.231
Output / 1M
$7.26
GLM 5.2 at roughly twice the throughput, for latency-sensitive workloads.
Agents resend their context every turn. With automatic prompt caching, repeated input bills at the cached rate. No configuration needed.
“Do you know what you pay for in agentic workloads? Cached tokens. A session with 50+ tool calls means the prompt is billed 50 times. All providers give a 1/5 cached discount for GLM 5.2. We at Fireworks dropped it to 1/10, matching GPT and Claude. That’s roughly 40% typical savings.”
You use what you use. You pay for what you use. Nothing pre-bought. Nothing wasted.
What will inference actually cost me?
We include inference with your package at cost. For example $5 with the starter package and $10 with the Pro package. Most personal agents run comfortably under $10 per month using DeepSeek Flash. Top up with a $10, $20, or $50 credit pack when your balance runs low. Credits don't expire. Prefer to use your own API keys? That works too just configure your own provider after setup.
Everything included. Nothing locked down
- Your VPS. Full root access, SSH in from day one.
- Hermes pre-installed. Running before you touch anything.
- Configure DeepSeek, GLM 5.2, MiMo UltraSpeed, or your own provider.
- Always on. Your agent runs 24/7.
- Real-time usage dashboard. See your balance and token spend at a glance.
- Low balance alerts. Notified before your agent pauses, not after.
- On-demand credit packs. $10, $20, or $50, no expiry, no auto-charge.
- One bill. Server and inference in a single monthly charge.
- Unlimited bandwidth. Host unlimited domains.
- Cancel any time. No lock-in, no annual contracts.
It's your server. We just set it up
Every plan is a real VPS. Not a container, not a shared environment. You get full root access and SSH credentials on sign-up.
Hermes is pre-installed and running when you arrive. Leave it exactly as it is and your agent is ready. Or SSH in, customise the config, add tools, run other software alongside it. It's your machine. Do what you want.
We handle the setup. You own the server.
Questions
Do I need to set up accounts with DeepSeek, Z.ai, or others?
You can use hermes.fast inference or configure your own provider during setup. Run hermes setup after SSHing into your server to choose the model provider you want.
Can I SSH into my server?
Yes. Full root access from day one. Install additional software, edit configs, customise Hermes. Nothing is locked down. It's your VPS.
How do I switch models?
Run hermes setup on your server, or use the Hermes web GUI under Settings → Model Provider.
How does inference billing work?
One credit balance, shared across all models. Your agent draws from it as it runs. When the balance runs low you get an email alert. Top up with a $10, $20, or $50 pack instantly. No auto-charges. Ever.
What happens if my balance hits zero?
Your agent pauses and you get a notification. Top up any time to resume. Credits never expire.
Which model should I use?
Start with DeepSeek V4 Flash. It's fast, cheap, and right for most Hermes tasks. Move to Pro or GLM 5.2 for deep reasoning or long-context engineering.
Can I upgrade my VPS plan?
Yes. Any time, from your dashboard. Your agent keeps running during the upgrade.
Is there a long-term contract?
No. Monthly billing. Cancel any time.
What does Solo include?
Solo includes a real VPS with Hermes pre-installed, monthly inference credit, and full root access from day one.
What is Hermes?
Hermes is an open-source AI agent built by Nous Research. It connects to messaging platforms, runs scheduled tasks, and acts as a persistent personal AI, running 24/7 on your own server.
Your agent. Running tonight
Pick a plan. SSH in. Everything's already running.
Full root access. Cancel any time.