.env.
This guide was written by Hermes Agent itself — it runs on ethereal and built
the tutorial by reading its own config.
1. Key
Hermes keeps secrets in~/.hermes/.env. Add your eth-... there:
- macOS / Linux
- Windows (PowerShell)
hermes config env-path.2. Provider
Open the config withhermes config edit (file ~/.hermes/config.yaml) and add
ethereal under providers:
| Field | Value | Why |
|---|---|---|
api | https://api.ethereal.llc/v1 | ethereal’s OpenAI-compatible endpoint |
key_env | ETHEREAL_API_KEY | name of the .env variable holding the key |
transport | openai_chat | request format (chat completions) |
3. Default model
model.provider points at the provider, model.default at the model id. You can
also do this without editing the file by hand:
4. Choosing a model
Open models (cheaper)
Besides Claude, ethereal serves open models (GLM, Qwen, Kimi) — much cheaper (12 / 60 ₽ per Mtok). Plug them in the same way, by id:
glm-5.2, glm-5.1, kimi-k2.7, qwen3.6, qwen3.7. The full
live list is at GET https://api.ethereal.llc/v1/models.
Charges follow the model’s rate for the actual tokens of the
session. See your balance and breakdown in your
dashboard.