OpenAI-compatible · pay per token, no subscription

Every frontier model.
One API.

Route requests to OpenAI, Anthropic, Google, DeepSeek, xAI and more through a single compatible endpoint. One key, one bill, pay only for the tokens you use.

Start building Browse models

Models: 59
Providers: 14
Max context: 1.1M
From: $0.015/1M

request.sh

curl https://platform.arx.ai.id/v1/chat/completions \
  -H "Authorization: Bearer $ARX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-sonnet-5","stream":true,
       "messages":[{"role":"user","content":"Explain serverless."}]}'

Drop-in replacement — same request shape, your OpenAI SDK untouched.

Serving models from the labs that matter

Catalogue

59 models. One endpoint.

Prices are per 1M tokens, billed per token. Switch models by changing a single string.

Full catalogue in dashboard →

Model Context Input / 1M Output / 1M Capabilities

Pricing

Pay for tokens. Nothing else.

No subscription, no seat fees, no minimum. Top up once and spend across every model.

Per-token billing. Charged on real input, cached and output tokens — no rounding up to a block.
One balance, all providers. A single credit pool works across every model in the catalogue.
Transparent rates. The price shown per 1M tokens is exactly what the dashboard bills.
No lock-in. Bring your own keys later or leave — your code stays OpenAI-compatible.

Prices below are pulled from the live catalogue. Cached input reads are billed separately and are usually 90% cheaper.

Cost estimator

Live rates

Model

Input tokens Output tokens

Estimated cost

$0.0000

—

Create an account

Estimates exclude discounts and cache reads.

Build

$0/month

Everything you need to ship your first agent.

All 59 models
OpenAI-compatible API
Streaming & tool calls
Per-token billing

Get started

Scale Pay as you go

Usage-based

For production traffic that needs headroom.

Higher rate limits
Per-key budgets & spend caps
Usage analytics per model
Prompt caching
Priority routing

Start scaling

Enterprise

Custom

Compliance, volume and support on your terms.

Volume pricing
Dedicated capacity
Invoicing & MSA
Private model access

Talk to us

Quickstart

Same SDK. New endpoint.

Point any OpenAI-compatible client at our base URL. Same request shape, same response shape, every model.

curl https://platform.arx.ai.id/v1/chat/completions \
  -H "Authorization: Bearer $ARX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "messages": [{"role": "user", "content": "Hello"}],
    "temperature": 0.7
  }'import os
from openai import OpenAI

client = OpenAI(
    base_url="https://platform.arx.ai.id/v1",
    api_key=os.environ["ARX_API_KEY"],
)

res = client.chat.completions.create(
    model="claude-sonnet-5",
    messages=[{"role": "user", "content": "Hello"}],
    stream=True,
)
for chunk in res:
    print(chunk.choices[0].delta.content or "", end="")import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://platform.arx.ai.id/v1",
  apiKey: process.env.ARX_API_KEY,
});

const stream = await client.chat.completions.create({
  model: "gpt-5.4-mini",
  messages: [{ role: "user", content: "Hello" }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

Base URL https://platform.arx.ai.id/v1 Pick a model →

OpenAI-compatible

Keep your SDK, prompts and tools. Only base_url and the model string change.

One bill

Prepaid credit shared across every provider. No per-vendor accounts to juggle.

Streaming & tools

Server-sent events, tool calls and JSON mode behave exactly as your SDK expects.

Prompt caching

Repeated context is billed at the cached rate where the provider supports it.

Model fallbacks

Swap providers during an outage by changing one parameter — no rewrite.

Keys you control

Per-key budgets, model allow-lists and expiry dates from the dashboard.

FAQ

Questions, answered.

Is it really OpenAI-compatible?

Yes. Chat completions, streaming, tool calls and embeddings follow the OpenAI schema, so the official SDKs and most frameworks work after changing the base URL to https://platform.arx.ai.id/v1.

How does billing work?

You prepay credit, and each request deducts the exact cost of its input, cached and output tokens at the published rate. Rates on this page are per 1M tokens; the dashboard bills per token.

Can I switch models mid-project?

Any time. The model is a string in the request body — change it and redeploy, or choose per request based on task difficulty and budget.

What about embeddings and vision?

The catalogue includes embedding models, and models with vision or reasoning capabilities are tagged in the table above and filterable.

Do you keep my prompts?

Requests are proxied to the upstream provider under their data policies and are not used to train models. Per-key analytics only store token counts and timing.

How do I get an API key?

Create an account, top up, then generate a key in the dashboard. Keys are shown once — store them in an environment variable.

Ship with every model, not one.

Create a key in under a minute. Start on the cheapest model, upgrade when it matters.

Get your API key Compare models

Every frontier model.One API.