Hermes Agent + Gonka AI — Autonomous Agent for Pennies

Hermes Agent is a self-learning open-source AI agent from Nous Research (MIT license). It's not just a chat: it's a full-fledged terminal CLI with memory, skills, a task planner, and a gateway to Telegram, Discord, Slack, and WhatsApp. The agent creates skills from experience, searches through past dialogues, and builds a user model between sessions. It can run on a $5 VPS or a GPU cluster.

The problem with Hermes is the same as with any agentic tool: token consumption. The agent reads files, runs tool-calling cycles, compresses context, and maintains memory — active work easily consumes tens of millions of tokens. With Anthropic or OpenAI models ($3-15 per 1M), this translates to tens and hundreds of dollars for a day of autonomous operation.

Hermes by default supports any provider: Nous Portal, OpenRouter, OpenAI, Anthropic — or your own endpoint. This is the entry point for JoinGonka Gateway: our OpenAI-compatible gateway provides inference to the decentralized Gonka network at a price starting from $0.0005 per 1M tokens — hundreds and thousands of times cheaper. Hermes becomes a truly everyday tool, not an expensive demonstration.

Step 1: Install Hermes and Get Your Key

Install Hermes (Linux, macOS, WSL2). The official installer will automatically pull Python, Node.js, ripgrep, and ffmpeg:

# Linux / macOS / WSL2
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

# Windows (PowerShell)
iex (irm https://hermes-agent.nousresearch.com/install.ps1)

After installation, restart your shell and check the launch:

source ~/.bashrc   # or ~/.zshrc
hermes             # interactive CLI

JoinGonka API key: If you don't have one yet — register at gate.joingonka.ai/register, get 10M free tokens, and create a key with the jg- prefix in the Dashboard. One key and one balance work for all models simultaneously.

Step 2: Connect Gonka as a custom provider

Hermes works with any OpenAI-compatible endpoint: if the server responds to /v1/chat/completions — Hermes can be directed to it. JoinGonka Gateway is exactly that. There are two setup methods.

Method 1: Setup wizard (recommended). Run in terminal outside an active session:

hermes model

In the list of providers, select "Custom endpoint (self-hosted / VLLM / etc.)" and enter three values:

API base URL: https://gate.joingonka.ai/v1
API key: jg-your-key
Model name: Qwen/Qwen3-235B-A22B-Instruct-2507-FP8

Method 2: Edit config. Hermes stores model settings in ~/.hermes/config.yaml — this is the single source of truth. Add a section:

# ~/.hermes/config.yaml
model:
  provider: custom
  default: Qwen/Qwen3-235B-A22B-Instruct-2507-FP8
  base_url: https://gate.joingonka.ai/v1
  api_key: jg-your-key

Verification: Launch hermes and type "Create a hello.py file with a hello world function." The agent should create the file and show the result. The current configuration state can be seen via hermes config show | grep '^model\.' and hermes status.

On-the-fly switching: Inside a session, use the command /model custom:Qwen/Qwen3-235B-A22B-Instruct-2507-FP8. And /model custom without a name will pull the model from the endpoint if there's only one.

Comparison of Autonomous Session Costs

Hermes is an agentic tool with memory and skills. It doesn't just respond with one reply: it reads files, writes code, executes commands, compresses context, and performs long background tasks through a planner. Each action is an API call, and tens of millions of tokens accumulate in a day of autonomous operation. Let's compare the cost of typical sessions:

Task	Tokens	Anthropic Claude	JoinGonka Gonka
One-time task with tool-calling	~5M	$15 — $75	$0.005
Background agent (several hours)	~20M	$60 — $300	$0.02
Autonomous feature development	~50M	$150 — $750	$0.05
A day of operation via gateway (Telegram + cron)	~150M	$450 — $2,250	$0.15

With JoinGonka Gateway, Hermes can be kept running 24/7: nightly reports, weekly audits, background tasks via a cron scheduler — without worrying about the bill. At Anthropic prices, each such launch would have to be manually calculated. 10M free tokens upon registration are enough for dozens of full sessions.

Model selection and tool calling

Through the Gateway, three Gonka models are immediately available to the user — all support native tool calling (OpenAI function calling), so Hermes skills and tools work reliably, without parsing text responses:

Model name (for config)	Context	Max. response	When to choose
`Qwen/Qwen3-235B-A22B-Instruct-2507-FP8`	128K	8192	Default — balance of quality and speed for agentic work
`moonshotai/Kimi-K2.6`	128K	3072	Strong reasoning and coding
`MiniMaxAI/MiniMax-M2.7`	128K	4096	Agentic scenarios, long tool-calling chains

About token limits in Hermes. In config.yaml, the context_length field is the full context window (input + output), and max_tokens is the ceiling for response length only. It's best not to set both: Hermes will automatically determine the correct values from the provider. If you set max_tokens manually, keep it within the model's limits — for Qwen3-235B, the maximum via Gateway is 8192. You can change the model at any time with the hermes model command or directly in the session via /model.

Similar terminal tools in our knowledge base: Claude Code, Cline.

Hermes Agent + Gonka = autonomous AI-agent from Nous Research for pennies. A day of background work costs $0.15 instead of $450-2,250 with Anthropic. Setup — custom provider with our gateway's base_url, native tool calling, three models to choose from. 10M free tokens are enough for dozens of sessions.

← LibreChat + Gonka AI — open-source ChatGPT Kilo Code + Gonka AI — AI-Agent in VS Code →

Want to learn more?

Explore other sections or start earning GNK right now.

Get 10M free tokens →