Inference Providers now route language models through OpenRouter. Each request goes to the best provider on live price, speed and reliability, and falls back to another one if it fails. Same code, same token. Add :fastest, :cheapest or a provider such as :novita to the model id to choose yourself.
On model pages such as zai-org/GLM-5.3, the widget has two new tabs: Providers lists each provider that can serve the model with its price, speed and uptime, and Metrics shows the best of each. You can turn off any provider in your settings. Custom provider keys no longer work for language models. More in the blog post.









