One API for leading AI models
Use selected chat, image, video, and audio models through an OpenAI-compatible gateway. Orux is designed to centralize routing, usage, and billing as each route is verified.
model: claude-sonnet-5
route: primary → fallback
status: ready
—
models available
100%
OpenAI-compatible
Pay per use
usage-based billing
Built-in
automatic failover
Every modality
Text, image, video and audio behind one key
The same base URL and the same key cover all four modalities. Long-running image, video and audio jobs go through the async task API and come back as hosted result URLs.
Chat & reasoning
Streaming SSE, tool calls and long context, on the OpenAI-compatible chat endpoint.
Image generation
Text to image and image editing, submitted as async jobs with hosted result URLs.
Video generation
Text and image to video, polled through the task API until the render finishes.
Audio & music
Music and speech generation, delivered through the same async task interface.
Sample outputs generated through Orux AI with the models listed in the model market.
Model market
Choose the right model for the job
Compare modalities and public usage prices without learning a different integration for every provider.
Integration
From API key to first response in three steps
Create an API key
Create a key from your workspace and set the limits your application needs.
Use the familiar SDK
Keep the OpenAI SDK and point its base URL to the Orux endpoint.
Switch by model alias
Change one model string while Orux selects an available upstream route.
A stable interface, not another SDK
The same authentication and request pattern works across all models. Async generation tasks use a consistent task lifecycle.
Read the quickstartcurl https://orux.top/api/v1/chat/completions \
-H "Authorization: Bearer sk-xxxxxxxx" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-5",
"messages": [{ "role": "user", "content": "Hello!" }]
}'Built for production
The operational details are part of the product
Routing is only useful when limits, failures, task state, and cost remain visible and predictable.
Multi-provider routing
Model aliases resolve to ordered routes with health-aware fallback.
Layered isolation
Key, application, and channel controls protect capacity under load.
Traceable billing
Usage records keep upstream cost and application charge separate.
Consistent async tasks
Image, video, and audio jobs share one observable task lifecycle.
FAQ
Common questions
Can I keep using the OpenAI SDK?+
Yes. Set the Orux base URL and API key, then select an available model alias. Provider-specific capabilities remain documented on each model page.
Which model types are supported?+
The gateway covers chat, image, video, and audio. Synchronous and asynchronous workloads use the endpoint pattern appropriate to each modality.
How is usage billed?+
Public prices are shown by the model's billing unit, such as tokens, image, second, or track. Your usage and charge records remain visible in the workspace.
What happens when an upstream provider fails?+
Eligible requests can move through the configured fallback chain. Circuit breakers and isolated connection pools prevent one provider from affecting the rest.
Build against one API, not a collection of vendors
Create a workspace, issue a scoped API key, and test your first model from the market.