Skip to main content

Divyam overview

Divyam gives your application frontier-model quality for less. You send the same Chat Completions requests you send today, to a Divyam endpoint, and name a Divyam model such as divyam-as/gpt-6-sol.

Why Divyam​

Each divyam-as/* model matches the quality of its equivalent frontier model at 50% of that model's price. The saving comes from Divyam's routing algorithm.

#1 on RouterArenaNov 2026

How routing works​

Your application asks for a catalog model. For each request, Divyam's routing picks a backing model that keeps that model's quality at a lower cost, and returns the answer in the usual Chat Completions format.

Your appCoding harnessDivyamrouterModel A62%Model B28%Model C10%
Your app or coding harness asks for one catalog model, and Divyam picks the backing model for each request.

The response's model field always shows the model you requested. To see which backing models answered, open Usage in the console and check Requested against routed. See Logs and usage.

Where to go next​