Hermes Agent gets a cloud instance with no model lock-in

Hermes Cloud is an optional hosted machine for Hermes Agent, the open-source agent from Nous Research. It is not the only place the agent can run. In its announcement post, Nous says Hermes Agent "has its own remote computer too, you just aren’t locked in to using it." The project's GitHub page shows what the other options are. The same agent can run on a $5 VPS, on a GPU cluster or on serverless infrastructure, using seven terminal backends that include local, Docker and SSH. It can also use models from Nous Portal, OpenRouter, OpenAI or a user's own endpoint, and switching models takes one command with "no code changes, no lock-in."

The Hermes Cloud page describes the hosted option as a dedicated, always-on instance with its own workspace and dashboard. When deploying, a user names the agent and picks a model and an instance size. The agent's files, memory and configuration stay in a persistent workspace whether the instance is running or stopped. Users can chat with it in the browser dashboard or connect messaging channels such as Telegram and Discord. Deploying requires $2 in credits or an active subscription.

The post says the hosted computer "costs a few dollars per month instead of $200." The post doesn't say what the $200 refers to, and only part of the thread could be collected, so that comparison can't be explained here. The FAQ on the cloud page shows what the instance price covers. Usage is charged to the user's Nous credit once a day, after the fact. A running instance pays for compute, metered to the second of uptime, plus storage. A stopped instance pays for storage only. Inference and tool usage are billed separately, on top of the instance price. The prices for each size are listed elsewhere in Nous Portal and were not in the material reviewed, so the few-dollar figure can't be checked against them. Even if it holds, it covers hosting only and leaves out the cost of the model calls and tools the agent uses.