01regna-1
Stop choosing
AI models.
regna-1 handles text, images, documents, and audio behind one name. It sends each request to a model that fits and fails over when one goes down.
OpenAI-compatible API · 1,000 free credits · No card
02Spec sheet
One name.
Everything behind it.
No per-model integrations, no model evaluations, no outage playbooks to keep. Call regna-1 and that's it.
- Input
- Text · images · PDF documents · audio (wav, mp3)
- Context
- Up to 1M tokens
- Output
- Streaming · tool calling
- Web search
- Add the web_search tool and the gateway runs the search, then answers from the results.
- Embeddings
- regna-embed. 1024 dimensions, picked on measured Korean retrieval quality.
- Failover
- On an error or timeout, the request moves to another model.
- API
- OpenAI-compatible (Chat Completions · Responses · Embeddings)
When a request comes in
- Textprimary text model
- Image · PDFmultimodal model
- Over 200K tokens1M-token model
- Audio · wav, mp3audio model
Regnax decides which model handles it. When the models change, your integration code stays the same.
03Metering
Metered, and visible.
Every response carries its input and output token counts. Credits are drawn for exactly those tokens, and the console shows usage by model and by key over 30 days. Calls that fail upstream without an answer are not charged.
- model
- regna-1
- usage.prompt_tokens
- 1,240
- usage.completion_tokens
- 312
Example
04Proof
Measured, not promised.
94.2%
RouterBench overall (n=400)
$0.01
average cost per question
0
API errors across the run
| Benchmark | regna-1 |
|---|---|
| MMLU (n=200) | 93.0% |
| ARC-Challenge (n=200) | 95.5% |
Measured end to end through the production API on 2026-07-13, routing overhead included.
Read the full methodology05Integration
Change one line.
Keep your stack.
Point your OpenAI SDK's base URL at regnax.ai. In the terminal, Regna Code works out of the box.
curl https://regnax.ai/v1/chat/completions \ -H "Authorization: Bearer $REGNA_API_KEY" \ -d '{ "model": "regna-1", "messages": [...] }'
from openai import OpenAI client = OpenAI(base_url="https://regnax.ai/v1", api_key=REGNA_API_KEY) reply = client.chat.completions.create(model="regna-1", messages=[...])
import OpenAI from "openai"; const client = new OpenAI({ baseURL: "https://regnax.ai/v1", apiKey: REGNA_API_KEY }); const reply = await client.chat.completions.create({ model: "regna-1", messages: [...] });
npm i -g @hyperez/regna-code06The math
Subscription stacks don't add up.
ChatGPT Plus, Claude Pro, and Gemini stacked side by side run about $60 a month, flat, whether you used them or not. Regnax works the other way: one API, metered by what each request actually consumed. Light month, light bill. Start free with 1,000 credits and read your real number off the console.
07Pricing
Simple pricing.
1 credit = $0.01 of usage, drawn only for what each request actually used. Sign up and get 1,000 credits, no card needed.
08FAQ
Questions, answered.
What is regna-1?
regna-1 is Regnax's default model name. It sends each request to a model that fits its type and length, fails over to another model during an outage, and returns a single response. Regnax evaluates and swaps the models behind it.
How is Regnax different from other AI gateways?
Instead of picking from a list of models, you call regna-1. Regnax evaluates and swaps the models. It is run by a Korean company, so you can settle with a Korean tax invoice.
Is the Regnax API OpenAI-compatible?
Yes. Point any OpenAI SDK or HTTP client at https://regnax.ai/v1 and use a Regnax API key with the model "regna-1". Chat Completions, Responses, and Embeddings are supported.
How does Regnax pricing work?
Usage is metered in credits: 1 credit equals $0.01 of usage, charged by what each request actually consumed. You start free with 1,000 credits, and paid plans start at $14 per month.
Do I need to change my code when new models are released?
No. Regnax evaluates new models and swaps them behind the regna-1 name. Your integration code and prompts stay as they are.
Can I get a Korean tax invoice?
Yes. Regnax is run by HyperEZ, Inc., a Korean company. Top up by bank transfer in KRW and we issue a Korean tax invoice. Card payments are processed in USD.