OpenAI-compatible access
Point an existing OpenAI-compatible Chat Completions client at the Arcvenza gateway while keeping the familiar message structure.
Gateway service available
Enterprise AI Model Gateway
Move an OpenAI-compatible chat client to Arcvenza with three settings: Base URL, SK, and model. Access is available to Arcvenza customers with a successful paid order.
OpenAI-compatiblePrivate SKChat Completions
Client protocolOpenAI-compatible
AccessPaid Arcvenza customers
CredentialPrivate SK
IntegrationBase URL · SK · Model
Core capabilities
A familiar client interface, managed through Arcvenza.
Configure an existing OpenAI-compatible client, choose from the models shown in your dashboard, and keep the SK lifecycle under your control.
Three request modes
Choose OpenAI Chat Completions, OpenAI Responses API, or Claude Code / Anthropic Messages. The dashboard switches the Base URL and request example with the selected mode.
Recommended models
Choose a model name returned in your dashboard. The current signed-in list is always the source of truth.
Personal SK management
Generate or regenerate your gateway SK from the signed-in console after account eligibility is confirmed.
AI Playground
Validate a model and prompt in the Arcvenza Playground before moving the same configuration into your own client.
Access workflow
From paid account to first request.
The signed-in console confirms eligibility and provides the values needed by your client.
Successful paid order requiredComplete an Arcvenza purchase
Gateway access becomes available after the account has at least one successful paid order.
Generate your SK
Open AI Model Gateway in the dashboard and generate the personal credential used by your clients.
Configure three values
Enter the displayed Base URL, your private SK, and a model name from the current dashboard list.
Choose the request mode
Use Chat Completions for OpenAI-compatible clients, Responses API for the newer OpenAI format, or Claude Code / Messages for Anthropic clients. The Base URL is different for Claude Code.
Integration
Make the first request.
Use the same Chat Completions payload structure from cURL or Python, replacing the placeholder with your own SK.
curl -X POST https://ai.arcvenza.com/v1/chat/completions \
-H "Authorization: Bearer sk-YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-5",
"messages": [
{"role": "user", "content": "Hello from Arcvenza"}
]
}'The examples use gpt-5 for illustration. Always use a model name from the current list in your signed-in dashboard.
Key safety
Treat the gateway SK as a private credential.
This public page never loads account data. Generate, view, and rotate credentials only in the signed-in console.
Regeneration invalidates the old SK
A regenerated SK takes effect immediately. Update every client that still uses the previous credential.
Keep it in your own client
Do not publish the SK or place it in URLs, client-side source, logs, analytics, or public repositories.
Use the matching interface
Keep the mode, Base URL, endpoint, authorization header, and request body together. Claude Code uses the Anthropic Messages format; OpenAI clients use Chat Completions or Responses.
FAQ
About the AI Model Gateway.
Who can use the AI Model Gateway?
Arcvenza customers with a successful paid order can access it. Sign in to the gateway dashboard to confirm eligibility and generate an SK.
Which model name should I use?
Choose a model listed in the signed-in dashboard. gpt-5 is only the example on this page; the current dashboard list is authoritative.
What happens when I regenerate my SK?
The old SK becomes invalid immediately. Replace it in every client before sending more requests.
What should I configure in my client?
Select a mode in the dashboard, then copy its Base URL, private SK, model, endpoint, authorization header, and request example. Claude Code uses /v1/messages after the Anthropic Base URL; OpenAI Chat Completions uses /chat/completions; Responses API uses /responses.
What should I check if a request fails?
Confirm paid access, the current SK, the displayed Base URL and endpoint, and a model from the current list. If the service remains unavailable, retry later or contact support.
Test before you ship