Nitro Variant

High-speed model inference with :nitro

The :nitro variant is an alias for sorting providers by throughput. When you use :nitro, OpenRouter will prioritize providers with the highest throughput (tokens per second).

Usage

Append :nitro to any model ID:

1{
2 "model": "openai/gpt-5.2:nitro"
3}

This is exactly equivalent to setting provider.sort to "throughput" in your request. For more details on provider sorting, see the Provider Routing documentation.