APIMaster.ai
APIMaster.ai Docs
中文Quick startAPIMaster.aiMarketplace

Hermes Desktop with APIMaster.ai

Set OpenAI-compatible API key and base URL in Hermes Desktop on Windows or macOS, then pick a model in Settings.

APIMaster.ai is OpenAI-compatible. Hermes reads two environment variables, then you pick the model in the desktop UI.

Quick setup

Step 1 — Get your API Key

  1. Open the APIMaster console.
  2. Copy your API Key (usually starts with sk-).
  3. Never paste a real key into chat or screenshots; use the placeholder your_apimaster_key below.

Step 2 — Edit the .env file

Windows

  1. Press Win + R, paste, and press Enter:

    %LOCALAPPDATA%\hermes

  2. Open .env in Notepad (create the file if missing).

  3. Typical full path:

    C:\Users\YourName\AppData\Local\hermes\.env

macOS

Hermes Desktop shares ~/.hermes with the CLI (official docs). Put keys in ~/.hermes/.env.

Option A — Finder (recommended)

  1. Go → Go to Folder… (Shift + Command + G).

  2. Paste:

    ~/.hermes

    (~ = your home folder; .hermes is hidden in Finder unless you use Go to Folder.)

  3. Open .env in TextEdit (create if missing).

Option B — Terminal

open ~/.hermes

Full path

/Users/YourName/.hermes/.env

Step 3 — Add these lines

OPENAI_API_KEY=your_apimaster_key
OPENAI_BASE_URL=https://apimaster.ai/v1
  • OPENAI_BASE_URL must end with /v1 — not https://apimaster.ai alone.
  • No spaces around =, no quotes unless Hermes docs say otherwise.
  • Save the file.

Step 4 — Restart Hermes Desktop

Fully quit Hermes (including the tray icon), then open it again.

Choose the model in the app

  1. Open Hermes Desktop.

  2. Go to Settings → Model → Main model.

  3. Provider: Custom endpoint.

  4. Model: pick a supported model id, for example:

    • claude-opus-4-8
    • claude-sonnet-4-6
    • gpt-5.5
    • gpt-5.4

    See the marketplace or your admin for the full list.

  5. Click Apply, then close Settings.

Hermes Desktop: Settings → Model → Main model

Test: pick a model and send a message

1. Set the main model in Settings (required)

  1. Open Settings (sidebar or bottom-left).
  2. Go to Model → Main model.
  3. Provider: Custom endpoint (or openai-api if shown).
  4. Open the Model dropdown and click an id (e.g. gpt-5.4) — do not type a display name.
  5. Click Apply and close Settings.

New chats use this model after Apply; old chats may still use the previous one.

2. Start a new chat

  1. Return to the chat list.
  2. Click New chat (or +).
  3. Use a fresh session, not an old thread.

3. Check the active model (recommended)

Near the input you may see something like openai-api / gpt-5.4. If it is wrong, fix Main model, Apply, and open another new chat. Some builds also offer a small model dropdown beside the composer.

4. Send a test message

Type Hello, reply with ok and press Enter. A normal reply means Key, base URL, and model are working.

5. Still failing?

Check error text (401, 404, timeout), confirm Hermes was restarted after editing .env, or run API connectivity test with the same key.

FAQ

Issue What to check
Send fails URL is https://apimaster.ai/v1; key copied fully; model id exact; Hermes restarted after .env edit
Folder missing Use ~/.hermes on Mac (not Application Support); launch Hermes once or mkdir -p ~/.hermes
No Custom endpoint Update Hermes to the latest version
Wrong models IDs come from APIMaster; see Supported models

Optional: API connectivity test with the same key.

Security

  • Do not share API keys in Slack, email, or public docs.
  • Use placeholders in tickets and screenshots only.
  • If leaked, revoke the key in the console, update .env, and restart Hermes.

Checklist

  • API Key copied from console (kept private)
  • OPENAI_API_KEY set in .env
  • OPENAI_BASE_URL=https://apimaster.ai/v1 (with /v1)
  • .env saved
  • Hermes fully restarted
  • Provider = Custom endpoint, model id selected, Apply clicked
  • Test message works in a new session

See also