Which AI provider key to use with Hermes Agent

Short answer. Hermes Agent does not include a model: it needs an account with an AI provider. On PlayNHost you paste an API key from OpenAI, Anthropic, Google Gemini, xAI or OpenRouter, and one key is enough. Hermes Agent picks the provider from the key, and you choose the model from the chat with /model. If you want many models behind one key and one bill, use OpenRouter. Your provider bills your model use directly; the server plan does not include it.

Checked on 7 November 2026, accurate for Hermes Agent v0.23.0.

This guide uses Hermes Agent's provider and configuration documentation, its source code for v0.23.0 where the docs are silent, and the PlayNHost panel as it works today.

The five keys the panel takes

An API key bills per use. A consumer subscription is not the same thing: Hermes Agent's docs say a Claude Pro plan cannot be used through its sign-in, and that the supported alternative is an Anthropic API key, billed per token.

Which one to choose

• You already pay one provider: use that key.
• You want to try several models: OpenRouter puts many providers behind one key and one bill. Hermes Agent's own docs list it as "recommended for flexibility".
• You are thinking of a free tier: Hermes Agent's docs warn that a free-tier Google Gemini key can run out after a handful of agent turns, because one message from you can take several model calls. Other small free quotas run out for the same reason.
• The model needs a large context. Hermes Agent requires at least 64,000 tokens of context and refuses smaller windows at startup. Check the context window of the model you pick, above all a small one.
• Not on PlayNHost: Nous Portal and other sign-ins (a ChatGPT, Claude Max or SuperGrok login) need hermes model or hermes setup on a command line, and local models (Ollama, LM Studio) need a model server the plan does not include. The panel takes API keys.

Add the key on PlayNHost

1. Open your server at panel.playnhost.com, then the Agent tab.
2. Under AI provider ("One key is enough. Each provider bills your own account for what your agent uses."), press Set on your provider's row, paste the key and press Save. The row then shows only the last four characters. The key is stored encrypted and never shown again; Replace and Remove change it later.
3. Add a Telegram or Discord bot token if you have not yet (Telegram, Discord), then start or restart the agent: "Changes apply the next time your agent starts."

The key reaches Hermes Agent through its environment, under the name in the table. Hermes Agent stores a fingerprint of it in auth.json, not the key itself (we checked this on a test server).

Choose the model from the chat

PlayNHost has no model field for Hermes Agent: you choose in the chat, with Hermes Agent's own command.

• Send /model to your bot. In Telegram it opens a picker: first the providers you have a key for, then their models. The current model is shown at the top.
• If you know the model's name, send /model <name>. Add --global to keep it for new conversations too.
• /model only switches between providers you have a key for. For another provider, add its key on the Agent page and restart.

Hermes Agent's docs add one cost note: switching models in the middle of a long conversation makes the next message re-read the whole conversation at full price, because the provider's prompt cache belongs to the old model. Switch early, or after /new.

Several keys at once

You can set more than one. When no provider has been chosen in Hermes Agent's settings, its v0.23.0 code takes an OpenRouter key first; without one, it takes the first provider with a key in its own built-in list. To avoid surprises, keep one key, or pick the provider with /model.

Keep the bill under control

• Your provider bills you, not us. PlayNHost does not resell model use or add it to your bill; the plan pays for the server.
• Set a spending limit or budget alert at your provider. Hermes Agent can run scheduled jobs on its own, and each run uses your key while you are not watching.
• Side tasks use your main model. By default Hermes Agent sends its side tasks (context compression, image reading, session titles, judging whether a command is risky) to your main model. Its docs note that with an expensive model these add meaningful cost.
• Only you can message it. The user ID lists on the Agent page keep strangers from spending your credits.

On your own machine

Run hermes model to pick a provider and model (it asks for the key or runs the provider's sign-in), or put the key in ~/.hermes/.env:

OPENROUTER_API_KEY=sk-or-...

Then restart the gateway. See how to run Hermes Agent 24/7, or create a Hermes Agent server and paste your key in the panel.

PlayNHost is not affiliated with Nous Research. Hermes Agent is open-source software under the MIT licence.