Great models.One connection.

Bring GPT, Claude and DeepSeek into the tools you love. Your keys, balance and usage. All in one place.

OpenAI-compatible API
Pay as you go
Usage at a glance

Interactive preview

import osfrom openai import OpenAI client = OpenAI(    base_url="https://codex.aicpr.icu/v1",    api_key=os.environ["PRISMROUTE_API_KEY"],) reply = client.chat.completions.create(    model="gpt-5",    messages=[        {"role": "user", "content": "Add retries to this request."},    ],)

import OpenAI from "openai"; const client = new OpenAI({  baseURL: "https://codex.aicpr.icu/v1",  apiKey: process.env.PRISMROUTE_API_KEY,}); const reply = await client.chat.completions.create({  model: "gpt-5",  messages: [    { role: "user", content: "Add retries to this request." },  ],});

curl https://codex.aicpr.icu/v1/chat/completions \  -H "Authorization: Bearer $PRISMROUTE_API_KEY" \  -H "Content-Type: application/json" \  -d '{    "model": "gpt-5",    "messages": [      {"role": "user", "content": "Add retries to this request."}    ]  }'

The only line you change

GPT / CodexFrom a first idea to your next release.

Done

Add retries to this request.

Sure. Retry with exponential backoff: wait 1, 2, then 4 seconds, try up to three times, and re-raise the last error if it still fails.

gpt-5Illustrative reply

Preview only. No API request is sent.Connection guide Explore GPT models

Simple to connect. Clear to run.

A clear path from your first connection to everyday work.

Your tools. Still yours.

Keep your SDK and client. Point the base URL at PrismRoute and add your key. That is the whole change.

client.py1 line changed

4 client = OpenAI(5-    base_url="https://api.openai.com/v1",5+    base_url="https://codex.aicpr.icu/v1",6     api_key=os.environ["PRISMROUTE_API_KEY"],7 )

OpenAI SDK
cURL
Python
JavaScript
Cherry Studio

See how to connect

One key. Many models.

GPT, Claude and DeepSeek behind one endpoint.

API keysk-••••3f2a

Switching takes one parametermodel="claude-sonnet-5"

All models

Know where it goes.

Balance and every call, with model, tokens and cost.

BalanceSample data$128.40

Model	Tokens	Cost
gpt-5	2,000	$0.0125
claude-sonnet-5	3,450	$0.0310
deepseek-chat	5,120	$0.0021

Open your workspace

Pay for what you use.

Input and output tokens are priced separately. The model directory shows each rate for your account group.

Input1,200 tokensInput rateSign in to view Output800 tokensOutput rateSign in to view

TotalThis request

Billing unit: As listed for each model
Your rate: Set by your account group
Every call: Recorded in your workspace

Browse the model directory

Your first connection. A few steps away.

Check your model, create a key, and make it part of your workflow.

01
Choose a model
Check its ID, endpoint and current rate.
02
Create a key
Set up access in your account console.
03
Connect your app
Follow the guide for your chosen API.

Request example

curl https://codex.aicpr.icu/v1/chat/completions \
  -H "Authorization: Bearer $PRISMROUTE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_MODEL_ID",
    "messages": [{"role": "user", "content": "Hello."}]
  }'

Replace the key and model ID with your own. The model directory lists supported endpoints.

A few thingsto know.

Where can I see models and pricing?

Sign in and open the model directory to check model IDs, compatible endpoints, billing units and rates for your account group. Your console shows actual balance and usage.

Can I keep using my existing tools?

Clients that accept a custom API base URL and key can use compatible models. For OpenAI-compatible clients, use https://codex.aicpr.icu/v1. Follow the documentation for your model’s endpoint.

How do I get help?

Email 321921230@qq.com with the request ID, time and a redacted error message. Keep API keys, passwords and private prompts to yourself.

Make room foryour next idea.

Create an account, add a key, and send your first request in minutes.

Get started Browse models