Token Supply

Setup guide

Configure Hermes with an OpenAI-compatible API

Yes. Hermes works with Token Supply. Keep Hermes's custom endpoint setup. Replace the base URL, set the context window, and store the Token Supply key in .env.

~/.hermes/config.yaml · ~/.hermes/.env

Create account Open docs

YAML, env, run

  1. Create an OpenAI-plan key

    Hermes is documented here for OpenAI-compatible plans.

  2. Edit config.yaml

    Set provider custom, base_url, chat_completions, and context_length.

  3. Set OPENAI_API_KEY

    Put the Token Supply key in ~/.hermes/.env. Do not commit it.

  4. Start Hermes

    If Hermes replies normally, check Usage in Token Supply.

Who this is for

Hermes users with a YAML model config who want an OpenAI-compatible host and a prepaid key.

Edit config.yaml

Open ~/.hermes/config.yaml and keep the custom endpoint setup. Set context_length because a custom endpoint often cannot auto-detect the window.

GPT-6.1 Sol, GPT-6 Sol, GPT-6 Luna, GPT-5.6 Sol, and GPT-5.5 use 1050000. Hermes chat_completions mode does not take a Codex-style model_reasoning_effort key; the example default stays gpt-5.6-sol.

~/.hermes/config.yaml
model:
  provider: custom
  default: gpt-5.6-sol
  base_url: https://api.token-supply.com/v1
  api_mode: chat_completions
  context_length: 1050000

Why this works: Hermes keeps its YAML model config. The base URL is the line that changes.

Edit .env

Open ~/.hermes/.env and replace the official provider key with your Token Supply key.

You can also run hermes config set OPENAI_API_KEY ts_live_... and let Hermes save it for you.

~/.hermes/.env
OPENAI_API_KEY=ts_live_...

Why this works: The key stays in .env so the YAML does not contain the secret.

  • Set context_length because a custom proxy often cannot auto-detect the window.
  • The model name is an example and stays gpt-5.6-sol. gpt-6.1-sol, gpt-6-sol, gpt-6-luna, and gpt-6.1-terra are also OpenAI ids on that key's plan.
  • You can also run hermes config set OPENAI_API_KEY ts_live_... .

Verify

Run hermes. If it replies normally, the setup is working. Check Usage in the Token Supply dashboard.

OpenAI SDK custom base URL · Claude Code for Claude plans

Details

Does Hermes work with Token Supply?
Yes. Hermes users with a YAML model config who want an OpenAI-compatible host and a prepaid key.
Where does the key go?
In ~/.hermes/.env as OPENAI_API_KEY. Do not put the key in config.yaml.
Why set context_length?
A custom endpoint often cannot auto-detect the window. GPT-5.6 Sol and GPT-5.5 use 1050000.
Can Hermes use a Claude plan?
This page covers OpenAI-compatible groups. Use Claude Code when you need the Claude Code client.

Related