JovethraDeveloper Cloud
Choose Starter
Jovethra / Hermes Agent

Run Hermes without being scared of the token counter.

Give your agent a metered OpenAI-compatible endpoint, a model name it already understands, and a monthly limit you can actually explain.

Connect Hermes in 2 minutes Estimate your usage
OpenAI-compatible endpointMonthly caps, no automatic overage10 accounts available
~/.hermes/config.yamlJOV / HERMES
model:
  provider: custom
  base_url: https://api.jovethra.xyz/v1
  api_key: ${OPENAI_API_KEY}
  default: gpt-5.6-sol

# ~/.hermes/.env
OPENAI_API_KEY=your_jovethra_key
Endpoint ready monthly usage metered
LIMITED FIRST COHORT10 accounts available

The first cohort is intentionally small while the service settles.

Reserve a Starter plan
01 / SETUP

From install to first agent turn in two minutes.

Hermes supports custom OpenAI-compatible endpoints. Keep the key in Hermes' environment file and the non-secret routing values in its YAML config.

01

Install Hermes

Choose the guide for your operating system.

Linux / macOS / WSL2 Run in Bash or Zsh, then reload the shell.
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash source ~/.bashrc
Windows PowerShell Run PowerShell as your normal user, then open a new window.
iex (irm https://hermes-agent.nousresearch.com/install.ps1)
02

Point it at Jovethra

Linux/macOS/WSL2: ~/.hermes/config.yaml
Windows PowerShell: $HOME\.hermes\config.yaml

model: provider: custom base_url: https://api.jovethra.xyz/v1 api_key: ${OPENAI_API_KEY} default: gpt-5.6-sol
03

Add your key and start

Save the key in the environment file for your platform, then launch a chat.

Linux / macOS / WSL2: ~/.hermes/.env
OPENAI_API_KEY=your_jovethra_key hermes
Windows PowerShell: $HOME\.hermes\.env Open a new PowerShell window after saving the file.
hermes

Linux/macOS/WSL2 use ~/.hermes/; Windows PowerShell uses $HOME\.hermes\. Never commit the key or place it in browser code. Read Hermes' custom endpoint guide

02 / AGENT USAGE

One agent turn is not one token.

Hermes can read files, run terminal commands, call tools, and continue a session. Model the whole workflow instead of guessing from the visible answer.

Practical rule: estimate repeated context as input, the visible answer as output, and cached system/tool context as cache.

EXAMPLE / 1 TASK
01
Read the repositorycontext + tool schemas
4.8k input
02
Run two commandsterminal calls + results
2.4k input
03
Write the responsereasoning and visible answer
800 output
04
Continue the sessionreused instructions and tools
12k cache
Estimated task footprint7.2k input · 800 output · 12k cache
03 / CALCULATOR

What would this usage cost?

Change the shape of a typical Hermes task. The calculator shows the smallest Jovethra plan that contains the monthly estimate.

10 tasks3,000 tasks
RECOMMENDED PLAN Starter Your estimate fits inside the Starter quota.
Input2.16M
Output240k
Cache3.6M
Monthly price€9.99
Choose Starter

Estimates are directional. Actual usage depends on prompts, tool output, session history, model behavior, and cache hit rates. Jovethra enforces the published input, output, and cache limits separately and does not bill automatic overage.

04 / WHY JOVETHRA

Make the agent budget legible.

Hermes remains the agent experience. Jovethra supplies the endpoint, the monthly boundary, and the usage view behind it.

Separate token categories

Input, output, and cache are visible as distinct quotas instead of one opaque counter.

No surprise overage

When a hard limit is reached, requests stop cleanly. A new paid period starts a new allowance.

One key, one account

Use the key issued to your Jovethra subscription and manage it from the customer portal.

Limited first cohort

Give Hermes a budget you can see.

10 accounts available while Jovethra validates the first Hermes workflows.

Start with StarterRead the API docs