Overview
One API key for many open models, compatible with the OpenAI and Anthropic APIs, billed per token in Rupiah.
The Helipod AI Gateway gives you one API key for many models: DeepSeek, GLM, Kimi, MiniMax, Hy3 and more. Helipod runs the gateway and routes your request to the model you name.
- OpenAI compatible.
POST /v1/chat/completionsworks with the official OpenAI SDKs and with any tool that lets you change the OpenAI base URL. - Anthropic compatible.
POST /v1/messagesworks with the official Anthropic SDKs and with Claude Code. - Responses API.
POST /v1/responsesworks with Codex and newer OpenAI clients. See Coding tools for Codex, Claude Code, opencode, Cline and others. - Every model in both formats. Pick the format your code already speaks. When a model natively speaks only the other format, the gateway converts the request and the answer for you.
- Pay per token, in Rupiah. Each request is charged by the tokens it used, from your AI balance. You top up the AI balance from your Helipod wallet. There is no subscription and no minimum spend.
Base URLs
| Format | Base URL | Endpoint |
|---|---|---|
| OpenAI | https://ai.helipod.io/v1 | POST /v1/chat/completions |
| OpenAI Responses | https://ai.helipod.io/v1 | POST /v1/responses |
| Anthropic | https://ai.helipod.io | POST /v1/messages |
The OpenAI SDKs expect the base URL to end in /v1. The Anthropic SDKs add /v1/messages themselves, so their base URL is the bare host.
Endpoints
/v1/chat/completions/v1/responses/v1/messages/v1/messages/count_tokens/v1/models/healthno keyok when the gateway is up.To read your balance, usage and AI keys or top up from code, use the Management API on api.helipod.io with a Helipod API key.
These are the only routes. There are no embeddings, legacy completions, Responses API, image or audio endpoints: any other path answers 404 with code not_found and a hint about the base URL.
How it works
- Your app sends a request with a Helipod AI key (
hel_ai_...). - The gateway checks the key, the model and your limits, then reserves the most the request could cost from your AI balance (a hold).
- The request goes to the model. The answer streams back to you as it is generated.
- When the answer is finished, the gateway reads the real token counts, charges that amount and releases the rest of the hold.
Requests that fail before or at the model are not charged. See Pricing and billing.
Prompts and answers pass through the gateway to produce the answer. Helipod stores only token counts, charges and timings for each request, never the text.
Quick start
Create an API key
In the dashboard, open a project and go to AI Gateway, then API keys. Create a key and copy it. Keep it on your server, for example in an environment variable:
export HELIPOD_AI_KEY="hel_ai_your_key"Keys belong to a project; the AI balance belongs to your account and is shared by all your projects. See Authentication.
Check or top up your AI balance
Open AI Gateway, then Overview. It shows your AI balance, any free credit, and what you spent this month. Top up from your wallet with a pack or a custom amount, or turn on auto top-up so your apps never stop. Without balance, requests answer 402 insufficient_balance.
Send your first request
curl https://ai.helipod.io/v1/chat/completions \
-H "Authorization: Bearer $HELIPOD_AI_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4.1-flash",
"messages": [{ "role": "user", "content": "Say hello in Indonesian." }]
}'The response is a normal OpenAI chat completion, with token counts in usage. The request and its charge appear under AI Gateway, Usage in the dashboard.
The dashboard also has a Playground to chat with any model, and a Quickstart with copy-ready setup for the OpenAI SDK, the Anthropic SDK and Claude Code.