OATHBREAKER / AGENT CONNECTIONS

Your agent.
Oathbreaker.

Run your agent with Oathbreaker.
One account, one API key, your own workspace.

Set up Hermes Already have Hermes?Use another agent

01 / RECOMMENDED

Hermes setup

OpenAI-compatible

Use the official Hermes app with Oathbreaker as its model provider. No GPU account or Oathbreaker-specific app is needed.

oathbreaker-glm-5.3Streaming and function calling. Prepaid usage at $3.00 input / $0.30 cached input / $5.00 output per million tokens. Model details.

Get your Oathbreaker key
Sign up with email or Google, add at least $5 in credit, then open API keys in your console. Create a key named Hermes and copy it when it appears. Existing customers can use their available credit.
Keep the key private. Paste it into Hermes when prompted; this page never asks for it.
Open your console Account, credit, and API keys
Install official Hermes
Choose your operating system in Nous Research’s installation guide. If Hermes is already installed, go straight to step 3.
Get Hermes Supported platforms
Connect the model
Exit any active Hermes conversation. In your terminal, run the command below and choose Custom endpoint. The wording may vary slightly by Hermes version.
hermes model
Run this in your terminal, outside an active Hermes conversation.
Official provider guide Use the connection details below
ENTER THESE DETAILS IN THE HERMES WIZARD
API base URL
https://oathbreaker.ai/v1
Include /v1 at the end.
API key
Your Oathbreaker key from step 1. Enter it in Hermes’ masked key prompt.
API compatibility mode
Chat Completions (chat_completions)
Model
oathbreaker-glm-5.3
Select this model if Hermes detects it automatically.
Context length
131072
The client context window; compression leaves room for tool results and output.
Display name
Oathbreaker
Save the setup, then apply these recommended settings in the same terminal. They set low reasoning and summarize long conversations earlier in this Hermes profile.
hermes config set agent.reasoning_effort low hermes config set compression.threshold 0.75 hermes config set compression.threshold_tokens 81920
These change only the named settings; they do not replace your config or API keys.
Oathbreaker allows up to 32,768 output tokens per request, including reasoning. Requests without an output limit use the API default; you do not need to edit YAML to get started.
Start a fresh conversation
Run hermes in your terminal, or open a new conversation in Hermes Desktop. Check that the selected model is oathbreaker-glm-5.3, then send this first prompt:
Reply with exactly: Oathbreaker connected.
Paste this into Hermes chat. The response is billed to your prepaid credit.
After the reply, your console’s Usage tab should show the request. Then try a small task in a folder you choose.
Check your usage New settings apply to new conversations

Need a hand?

My terminal cannot find the hermes command

Open a new terminal after installation. If the command is still unavailable, follow the command-line instructions for your operating system in the official installation guide.

Hermes still uses my previous provider

Run hermes model again and select your saved Oathbreaker endpoint. Start a new conversation rather than resuming an old one.

The key is rejected or credit is missing

Use an active Oathbreaker API key, not your Google login or another provider’s key. Check available credit in your console. If you lost the one-time key, create a replacement there and update Hermes.

A long conversation or tool request fails

Start a fresh conversation or ask Hermes to summarize the current one. Keep the recommended context settings above. Oathbreaker accepts up to 128 tool definitions per request. See API limits and errors.

Where do my files and prompts go?

Hermes runs your workspace and tools on your computer unless you configure remote tools. Prompts, conversation history, and any file or tool content included in a request are sent to Oathbreaker and its inference provider. External tools may use their own services. Read our privacy policy.

Still stuck? Contact support with the error message. Never send your API key.

02 / MORE CONNECTIONS

Use another agent

Start with credit and an API key from your console. Choose your client below and keep the key in its local settings or your terminal environment.

Set your API key for terminal agents

In Bash or Zsh, run this, paste your key at the hidden prompt, and press Enter. Start the agent from the same terminal.

printf 'Oathbreaker API key: '; read -rs OATHBREAKER_API_KEY; export OATHBREAKER_API_KEY; printf '\n'

The key stays in this terminal session.

In PowerShell 7.2 or later, use the masked prompt below.

$env:OATHBREAKER_API_KEY = Read-Host -MaskInput "Oathbreaker API key"

Start the agent from this PowerShell session.

OpenCodeRepository coding agent

Install OpenCode and set OATHBREAKER_API_KEY in your terminal.
In your workspace, create opencode.jsonc with the config for your version. If you have a config already, merge the Oathbreaker entry into it. Run opencode --version to check your version.

{
  "provider": {
    "oathbreaker": {
      "name": "Oathbreaker",
      "npm": "@ai-sdk/openai-compatible",
      "options": {
        "baseURL": "https://oathbreaker.ai/v1",
        "apiKey": "{env:OATHBREAKER_API_KEY}"
      },
      "models": {
        "oathbreaker-glm-5.3": {
          "name": "oathbreaker-glm-5.3",
          "tool_call": true,
          "limit": {
            "context": 131072,
            "output": 32768
          }
        }
      }
    }
  }
}

Use this config for OpenCode 1.x.

Config for OpenCode 2.x

{
  "providers": {
    "oathbreaker": {
      "name": "Oathbreaker",
      "package": "@opencode/ai/providers/openai-compatible",
      "settings": {
        "baseURL": "https://oathbreaker.ai/v1",
        "apiKey": "{env:OATHBREAKER_API_KEY}"
      },
      "models": {
        "oathbreaker-glm-5.3": {
          "name": "oathbreaker-glm-5.3",
          "capabilities": {
            "tools": true,
            "input": [
              "text"
            ],
            "output": [
              "text"
            ]
          },
          "limit": {
            "context": 131072,
            "output": 32768
          }
        }
      }
    }
  }
}

Use this config for OpenCode 2.x.

From that folder, start OpenCode with this model.

opencode --model oathbreaker/oathbreaker-glm-5.3

You can also select the model from /models.

OpenWorkDesktop workspace

Get OpenWork and open a local workspace. Add the config below to opencode.jsonc in that folder, merging the provider if a config exists.
In Settings → Environment, add OATHBREAKER_API_KEY with your key as its value and apply the change.
Run Reload OpenCode config from the command palette or restart OpenWork. Choose oathbreaker-glm-5.3 in the chat model picker.

{
  "provider": {
    "oathbreaker": {
      "name": "Oathbreaker",
      "npm": "@ai-sdk/openai-compatible",
      "options": {
        "baseURL": "https://oathbreaker.ai/v1",
        "apiKey": "{env:OATHBREAKER_API_KEY}"
      },
      "models": {
        "oathbreaker-glm-5.3": {
          "name": "oathbreaker-glm-5.3",
          "tool_call": true,
          "limit": {
            "context": 131072,
            "output": 32768
          }
        }
      }
    }
  }
}

The workspace file references your locally stored key.

OpenWork’s custom provider guide

PiLightweight terminal agent

Install Pi and set OATHBREAKER_API_KEY in your terminal.
Create ~/.pi/agent/models.json with the config below, or merge the Oathbreaker entry into its existing providers.
Run pi from the same terminal, then use /model to select oathbreaker-glm-5.3.

{
  "providers": {
    "oathbreaker": {
      "baseUrl": "https://oathbreaker.ai/v1",
      "api": "openai-completions",
      "apiKey": "$OATHBREAKER_API_KEY",
      "compat": {
        "supportsStore": false,
        "supportsDeveloperRole": false,
        "maxTokensField": "max_tokens"
      },
      "models": [
        {
          "id": "oathbreaker-glm-5.3",
          "name": "oathbreaker-glm-5.3",
          "contextWindow": 131072,
          "maxTokens": 32768
        }
      ]
    }
  }
}

Pi reads the key from the terminal environment.

Pi’s compatible endpoint guide

Send “Reply with exactly: Oathbreaker connected.” Then check Usage in your console for the request.

Hermes is an open-source project by Nous Research. Oathbreaker provides independent model inference for the clients listed above.

Hermes setup

Get your Oathbreaker key

Install official Hermes

Connect the model

Start a fresh conversation

Need a hand?

Use another agent