Hormouz AIHormouzSign in

Fast, reliable AI with no outages

When your product depends on an AI model, an outage at that model takes your product down too. With Hormouz AI, if a provider or network route goes down, your request goes over another route automatically and still gets an answer, and you do not change your code.

Create an account

Why AI services go down in Iran

Filtering and limits on international internet, sanctions, outages at the providers, rate limits and peak-hour congestion can each take a direct key or a simple reseller offline for hours.

Reliable AI: how Hormouz keeps working

Hormouz has several providers and several network routes for its models and keeps measuring the health of each route. If one slows down or stops answering, the next request, and even the same one, moves to a healthy route. The response has the same OpenAI-compatible format, and your app sees no difference.

Hormouz runs inside Iran and has backup routes for the days when international internet is disrupted. You call api.hormouz.net, and Hormouz picks the route, retries on errors and switches routes when it has to.

The fastest AI API for your workload

Response speed depends on the model and on the route a request takes. Hormouz picks the fastest healthy route for each request and streams the answer token by token. Prompt caching also cuts the cost and time of repeated text.

For work where speed matters most, you will find light models such as Gemini Flash, GPT mini and DeepSeek on the models page, and switch between them by changing one word in the request.

For products that cannot go down

Teams such as Quera, Hooshang, Edna and Karafs process millions of tokens through Hormouz every day, and as of October 2026 Hormouz has answered more than 45 million requests. The dashboard shows the cost, model and response time of every request, and support is in Persian.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.hormouz.net/v1",
    api_key="ng-…",
)

FAQ

What happens if a model or provider goes down?
Your request moves to a backup route automatically and still gets an answer; you do not need to change your code.
Which is the fastest AI API?
Speed depends on the model and the route. Hormouz picks the fastest healthy route for each request and streams the answer; for the most speed, choose a light model such as Gemini Flash or GPT mini.
Do I need to change anything in my code?
Only the base URL, once, to api.hormouz.net/v1. Switching between routes is automatic.
Does it work when international internet is disrupted?
Yes. Hormouz runs inside Iran and has backup routes for those days.
Which tools does Hormouz work with?
Any tool that accepts an OpenAI-compatible API, such as the OpenAI SDK, LangChain, the Vercel AI SDK, and the VS Code and Cursor editors. Claude Code and Gemini CLI only speak Anthropic's and Google's own APIs and do not connect.

Hormouz AI: one fast, reliable API for every AI model.