1. Step 1: Add RouterHub as a custom endpoint

    Outside an active chat, run:

    Open the provider wizard
    hermes model

    Choose Custom endpoint (self-hosted / VLLM / etc.) and enter:

    SettingValue
    API base URLhttps://routerhub.si/v1
    API keyYour RouterHub API key
    API mode, if askedchat_completions
    Model nameCopy a Chat Completions model ID from the models and pricing page
    Context length, if askedUse that model’s current limit from the models and pricing page

    Choose a model with at least 64K context. The wizard saves your settings automatically.

    macOS / Linux

    Saved in ~/.hermes/config.yaml.

    Windows

    Saved in %USERPROFILE%/.hermes/config.yaml.

  2. Step 2: Set request timers and start Hermes

    Before launching Hermes, set the ten-minute API and stale-response limits in this terminal. Hermes uses seconds here, not milliseconds:

    macOS / Linux

    macOS / Linux
    export HERMES_API_TIMEOUT=600
export HERMES_API_CALL_STALE_TIMEOUT=600
export HERMES_STREAM_STALE_TIMEOUT=600
export HERMES_STREAM_READ_TIMEOUT=600
hermes

    Windows (PowerShell)

    Windows (PowerShell)
    $env:HERMES_API_TIMEOUT = "600"
$env:HERMES_API_CALL_STALE_TIMEOUT = "600"
$env:HERMES_STREAM_STALE_TIMEOUT = "600"
$env:HERMES_STREAM_READ_TIMEOUT = "600"
hermes

    These affect this terminal session, not sandboxed tool execution. Existing provider/model request_timeout_seconds, timeout_seconds, or stale_timeout_seconds overrides take precedence; merge or adjust those entries rather than duplicating the provider. For Desktop, use those provider settings in the wizard-saved ~/.hermes/config.yaml; terminal environment changes do not update an already-running Desktop app. To change models later, exit the chat and run hermes model again.

Need to install Hermes Agent?

Get Hermes Desktop on macOS/Windows (CLI included), or follow the Linux quickstart. Open a new terminal after installing.

Check the connection (optional)

When you’re ready to use credits, open an empty project and send this short test:

Reply with OK. Do not read files, run commands, or make edits.

Open RouterHub usage and check that the request appears under the right model. Keep auto-approval off while testing. A text reply checks the connection, not all coding features.

Troubleshooting

A request or stream times out

Confirm the four HERMES_*TIMEOUT values are 600 seconds in the terminal used to launch Hermes. Check for provider/model overrides in ~/.hermes/config.yaml. A stale timer measures a silent pause, not the full task duration. Check Usage before retrying; do not disable sandbox or terminal deadlines to change API timeouts.

I only see my old provider

Exit the chat and run hermes model. In-chat /model only switches between configured providers.

Invalid key, model, or URL

Reopen the wizard. Check your RouterHub key, https://routerhub.si/v1, chat_completions, and the exact catalogue ID. Do not add an OpenRouter author prefix.

Context or tool errors

Recheck the selected model’s current limits. Run hermes doctor for local setup issues; redact keys and private paths before sharing logs.

Still stuck? Contact support.

Common questions

Which models should I choose?

Use a Chat Completions model with a published context window of at least 64K. Check the current models page; if the limit is missing, confirm it with support. A missing limit is not proof of incompatibility.

Does this cover Desktop too?

The official installer includes both interfaces. These steps use the CLI wizard; neither interface has been tested end to end with RouterHub. Other tools or providers can have separate charges.