One request body
Chat, image, audio, and embeddings share the same fields. Swap the model id when you want a different one.
Priced in rupees
Call any modelfrom one API
OpenAI, Anthropic, Google, and the rest of the catalog. Text, image, audio, and video. Change the model name. The request stays the same.
Providers on this gateway
OpenAI
Anthropic
Google
Mistral
DeepSeek
Qwen
How a call moves.
You post once. Name the model, or leave it on auto. The reply uses the same fields either way.
New request
Request type
ChatEmbeddings
Model
auto
Routing
balanced
Message
Summarize the deploy notes.
CancelSend
Four jobs, one gateway
A catalog you can read
Text, reasoning, vision, image, audio, and video, with context length and a price in rupees.
Routing, if you want it
Cheaper, faster, or a model you named. The HTTP call does not change.
The next model, if this one fails
A timeout or a provider error moves the request to the next model you allowed.
Worth a look before you choose
Usage
Requests, tokens, latency, and spend, listed by model.
Fallback
The next model on your list takes the call when the first one fails.
Side by side
Context length, what the model accepts, and the rupee rate.
Same JSON back
The response fields do not change when the provider does.
A few languages
JavaScript, Python, PHP, and Go. Underneath, it is the same HTTP call.
The price first
Each model page shows a rupee rate before you pick a route.
Open the catalog. Ship on one API.
- 1
Pick from the catalog
Or skip the choice and let a policy pick by price or speed.
- 2
Point the client at one URL
Chat, image, and audio use that same body.
- 3
Send a sample
You can see the request, the reply, and the token count on the model page.
- 4
Leave the client alone
Later you change the model field. You do not rewrite the integration.
Find a model.Send one request.
Chat, image, audio, and video are listed with a price in rupees.