Skip to main content
QwenCloud home page
Docs
Clients & tools

Hermes Agent

Terminal AI coding assistant by Nous Research

Hermes Agent is a terminal AI coding tool that connects to QwenCloud through pay-as-you-go, Token Plan (Personal Edition), or Token Plan (Team Edition).

Install Hermes Agent

  1. Run this command to install Hermes Agent. The script automatically installs dependencies like Python and Git.
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
Windows does not support native installation. Install WSL2 first, then run the above command in WSL2.
  1. After installation, reload the terminal environment.
source ~/.bashrc    # If using zsh, change to source ~/.zshrc
  1. Verify the installation. A version number confirms success.
hermes --version

Configure access credentials

Run hermes config set to configure the Base URL and API Key for your chosen plan:
  • Token Plan (Personal Edition): Individual subscription; token consumption deducts Credits.
  • Token Plan (Team Edition): Seat-based subscription; token consumption deducts Credits.
  • Pay-as-you-go: Post-paid based on actual usage.
The examples in this topic use the Anthropic-compatible protocol: the Base URL ends with /apps/anthropic and api_mode is set to anthropic_messages. Hermes Agent also supports the OpenAI-compatible protocol: replace the trailing /apps/anthropic in the Base URL with /compatible-mode/v1 and remove the api_mode setting. For example, the OpenAI-compatible Base URL for Pay-as-you-go (Singapore) is https://maas.qwencloudapi.com/compatible-mode/v1.In addition to the command-line version, Hermes Agent also offers a desktop version (Hermes Desktop). You can download the installer from the Hermes website, or run hermes desktop after installing the command-line version. Hermes Desktop and the command-line version share the same ~/.hermes/config.yaml configuration file, so the access parameters are identical to those described here. When connecting through a Custom Endpoint in the desktop app, use the OpenAI-compatible Base URL described above.

Token Plan (Personal Edition)

Replace YOUR_API_KEY with your Token Plan (Personal Edition) API Key. Available models are listed in supported models. 
hermes config set model.provider custom
hermes config set model.base_url https://token-plan.maas.qwencloudapi.com/apps/anthropic
hermes config set model.api_mode anthropic_messages
hermes config set model.api_key YOUR_API_KEY
hermes config set model.default auto
Important: qwen3.8-max thinking mode (hybrid — can be enabled/disabled):
  • temperature: Defaults to 0.6 in thinking mode. Values below 0.6 are automatically adjusted to 0.6.
  • reasoning_effort: Controls reasoning depth. Options: xhigh, medium, low. Default: xhigh.

Token Plan (Team Edition)

Replace YOUR_API_KEY with your Token Plan (Team Edition) API Key. Available models are listed in supported models. 
hermes config set model.provider custom
hermes config set model.base_url https://token-plan.maas.qwencloudapi.com/apps/anthropic
hermes config set model.api_mode anthropic_messages
hermes config set model.api_key YOUR_API_KEY
hermes config set model.default auto
Important: qwen3.8-max thinking mode (hybrid — can be enabled/disabled):
  • temperature: Defaults to 0.6 in thinking mode. Values below 0.6 are automatically adjusted to 0.6.
  • reasoning_effort: Controls reasoning depth. Options: xhigh, medium, low. Default: xhigh.

Pay-as-you-go

Replace YOUR_API_KEY with your QwenCloud API Key. Available models are listed in Anthropic compatible API. Set base_url for your region. The API Key must match the selected region:
  • Singapore: https://maas.qwencloudapi.com/apps/anthropic
hermes config set model.provider alibaba
hermes config set model.base_url https://maas.qwencloudapi.com/apps/anthropic
hermes config set model.api_mode anthropic_messages
hermes config set model.api_key YOUR_API_KEY
hermes config set model.default qwen3.7-max

Verify configuration

Send a test message to verify your configuration: 
hermes chat -q "Hello"
A successful response confirms the configuration works. To switch models, use the -m parameter: 
hermes chat -m qwen3.7-max

FAQ

Error codes

For configuration errors, check the FAQ for your billing plan: