Skip to content

Hermes Agent with Keln

Run Hermes Agent on Keln models by installing the Keln provider plugin.

Before you start

  • Hermes Agent 0.22.6 or newer.
  • A Keln API key. Sign up at app.keln.ai, then open Keys → Create key.

Connect Hermes Agent to Keln

1. Install the Keln provider

hermes plugins install keln-provider

Enter the following when prompted:

Hermes asks for Enter
KELN_API_KEY sk-keln-… your key
Enable 'keln-provider' now? y

Hermes saves the key in ~/.hermes/.env.

If Hermes did not ask for the key, or you skipped it, add it with:

hermes config set KELN_API_KEY sk-keln-…

2. Choose Keln

hermes model

Enter the following when prompted:

Hermes asks for Enter
Provider Keln
Keln API key: [K]eep / [R]eplace / [C]lear Press Enter to keep your key
Base URL Press Enter to keep https://api.keln.ai/v1
Model A chat model, for example moonshotai/kimi-k3

Don't skip this step. Installing the plugin selects Keln as the provider, but Hermes keeps its default model until you choose a Keln model here.

3. Start a chat

In Terminal:

hermes chat

Or open the Hermes app and start a new chat. The model menu next to the message box shows a short list under KELN. To find another Keln model, type its name in Search models, for example glm. Click ☆ next to a model to keep it at the top of the menu.

Chats that were already open keep their old model. Start a new chat after you switch.

Add credit under Billing → Add credit before you start chatting. Keln returns a 402 error if your balance is empty.

Models

Hermes Agent works with all Keln chat models. See Models & pricing for the full list and pricing.

Hermes also lists other Keln models for speech, transcription, embeddings, reranking, and decisions. These can't be used as the agent's main model.

Want to pay less? Open the API keys page on Keln and set your key's mode to Lite. You'll get the same model at a lower price, though it may be slower than Standard.

Learn more about Standard and Lite modes.

Usage

Your Keln Usage page shows your spend, tokens, latency, and error rate. Keln stores usage metadata, not your prompts or responses (ZDR).

The Keln Usage page with sample data: spend, requests, tokens, time to first token, error rate and a spend-by-day chart

Switch models

Run hermes model to change the default model for new chats.

To switch only the chat you're in, type:

/model zai-org/glm-5.3 --provider keln

Where to go next

Try Keln

Get a Keln API key to start using open models in Hermes Agent.

Get an API key Models & pricing ↗