Hostinger Hermes Agent (2026): Pricing, Setup & Honest Review

What is Hermes Agent?

Hermes Agent is an open-source, self-improving AI agent developed by Nous Research. That distinction matters: Hostinger did not build Hermes — it hosts and sells it as a managed app and as a one-click VPS template. The software itself is free under the MIT license; what you pay for is hosting plus the model and credit usage. Hostinger describes it as an agent “that can execute tasks, learn from interactions, and expand its capabilities over time.”

In practical terms, Hermes is aimed at autonomous work: it can execute complex tasks, run code when needed, learn from what it does, and expand its own capabilities over time. Think of it less as a chatbot you talk to and more as a worker you hand jobs to.

What exactly does Hostinger offer?

Hostinger offers Hermes through two distinct routes:

• Managed — “Managed AI Automations” plan: one subscription covers four apps — Hermes, OpenClaw, n8n, and Paperclip. You pick one app to start and can switch between apps at any time under the same subscription (switching starts the new app fresh; configurations and history do not transfer). Hostinger lists: launch in about 60 seconds, no technical setup, long-term memory, a visual interface plus CLI, Telegram and WhatsApp connection, the option to use your ChatGPT Plus/Pro subscription, pre-configured AI credits, and security managed by Hostinger.
• Self-hosted — VPS Docker template: a one-click Hermes Agent template in the VPS Docker Manager Application Catalog. It deploys a Docker container (plus Traefik), with CLI access via the “Open” button in the VPS dashboard.

Pricing: the honest version

The managed plan’s headline pricing is below, rendered from our deals table. Hostinger also sells 12-month and month-to-month terms at higher rates, and the plan renews at the standard renewal price. Now the honest part: the managed plan includes initial AI credits via the Hostinger AI Router (nexos.ai) for Claude, Gemini, and GPT models, plus web search via Oxylabs — but heavier usage may require additional credit top-ups. The headline price is a starting price, not a guaranteed total. We say this plainly because most coverage of this plan doesn’t.

The self-hosted route uses standard KVM VPS pricing. Hostinger’s own tutorial recommends at least 2 vCPU and 8 GB RAM for a stable Docker deployment (KVM 2); KVM 1 can run Hermes with cloud models according to third-party guides, but Hostinger’s recommendation is the safer bet. Hostinger’s cost tutorial puts a typical self-hosted total anywhere from a few dollars a month to well over eighty — the range is wide because it depends entirely on how much model usage you burn.

One more honest note: we could not find a published USD price for additional AI-credit top-up packs for the managed plan, so we are not printing one. Check the live cart or your dashboard before committing to heavy usage.

How AI credits work (and what they cost)

Both routes can use Hostinger’s nexos.ai credit system — the same shared credit pool described in Hostinger’s OpenClaw credit guides. Example costs per one million tokens: Gemini 3 Flash 0.5 in / 3 out credits; GPT-4.1 2.2 / 8.8; Grok 4 3 / 15; Claude Sonnet 4.5 3.3 / 16.5; Claude Opus 4.6 5.5 / 27.5; Claude Opus 4 15 / 75. The credit balance is shared across Hostinger AI products that use AI credits, and it can never go below zero — requests are simply declined if the balance is too low.

What happens if your credits run out? On the self-hosted route, Hermes stops generating responses until you top up — you check and top up at Docker Manager → Projects. On the managed plan, you top up from the Hostinger dashboard. If you’d rather avoid per-token billing entirely, you can connect your ChatGPT Plus/Pro subscription via OAuth, or paste your own OpenAI/Anthropic API key. (Separately, Nous Research itself sells an optional “Nous Portal” subscription for self-hosted users — that’s Nous’s product, not Hostinger’s, starting with a free pay-as-you-go tier.)

Setup: managed route

Per Hostinger’s managed product page (read October 9, 2026): pick the app (Hermes), connect a messaging channel and an AI model, and follow the guided onboarding — Hostinger says about 60 seconds, no coding needed. From there, everything is managed: server configuration, updates, backups, and security.

Setup: self-hosted VPS route

Per Hostinger’s official Hermes-on-VPS setup guide, the steps are:

1. Get a KVM VPS — Hostinger’s tutorial recommends at least 2 vCPU and 8 GB RAM.
2. VPS dashboard → Docker Manager → Catalog → search “Hermes Agent” → Select.
3. Set an admin username and password (keep them safe).
4. Optionally connect Hostinger AI Router credits (auto-filled if purchased with the VPS) and/or Oxylabs credits for web search and scraping (auto-filled if purchased; otherwise enter your own API keys manually).
5. Click Deploy. Hostinger creates and starts the Hermes container plus services like Traefik.
6. Access it via the “Open” button next to the Hermes project — this opens the Hermes CLI.
7. Complete the first-run CLI setup flow (arrow keys + Enter/Space): choose Full setup (your own providers, including Hostinger AI/Oxylabs credits) or Quick Setup (Nous Portal sign-in, no API keys); select a model provider (the Hostinger AI Router key is pre-filled if purchased); toggle communication channels (Telegram, WhatsApp, and others); and enable tools — check Web Search & Scraping and select Oxylabs as the provider if you purchased credits.

Type /help in the CLI for the available commands. As with any self-hosted setup, updates, security, and recovery are your responsibility from here on.

Hermes vs OpenClaw: Hostinger’s own distinction

Since one subscription covers both, which should you actually run? Hostinger’s product-page FAQ draws the line this way:

• Managed OpenClaw = a personal AI assistant for conversational automation through messaging channels — the agent you chat with on WhatsApp or Telegram.
• Managed Hermes Agent = a self-improving agent built for task execution and autonomous workflows — the agent you hand jobs to, which learns, executes complex tasks, runs code when needed, and expands its capabilities over time.

Our read: pick OpenClaw if your main use case is conversational — an assistant living in your messaging apps. Pick Hermes if your main use case is work — tasks that run, iterate, and improve. And remember that switching between them under the managed plan starts fresh each time, so commit to one for any serious project. Our full OpenClaw guide covers the other side in depth.

No relation to Horizons or AI Builder

One common confusion to clear up: Hermes has no relation to Hostinger Horizons or the AI Website Builder. Per Hostinger’s August 2026 announcement, Horizons and the classic Website Builder were merged into a single product called AI Builder — that’s for building websites and web apps. Hermes is a separate product line entirely: autonomous task-execution agents. They solve different problems and are billed separately.

Who should pick which route?

Pick the managed plan if: you want an agent running today with zero server work, you like the flexibility of switching between Hermes, OpenClaw, n8n, and Paperclip under one subscription, and you’re comfortable with usage-based credit costs on top of the base price.

Pick the VPS route if: you want full control over the environment, you already run a Hostinger VPS, you prefer to bring your own API keys and pay model providers directly, or you want to run Hermes alongside other Docker workloads. Size it at 2 vCPU / 8 GB RAM per Hostinger’s own recommendation.

FAQ

Comparing your options? Read our OpenClaw guide for the conversational-assistant side, n8n hosting for workflow automation on the same managed plan, or Hostinger VPS plans if you’re going the self-hosted route.