Inference provider request

[winter jungle]

I recently submitted an application to become an inference provider and wanted to check if there’s a way to expedite the review process. Our infrastructure and API endpoints are ready, and we’re eager to start contributing to the network.
Could someone please let me know the usual review timeline or if any additional info is needed from our side?

[rough sundial]

The wait is months long unless you offer something unique no other providers (or very few other providers) have like super fast speeds or uncommon models or prompt caching for all models or something else

[winter jungle]

We’ve built a prompt caching layer and are continuously improving it, our performance is already very fast. For the uncommon models, we currently have 2.2 million models on our platform ready to be deployed. Based on the uncommon models you mentioned, could you share some examples of the model types you’re looking for? eg. Computer Vision, Object Detection, Custom Embedding etc…

Also, could you share your expectations regarding throughput?

[rough sundial]

I'm not on the OpenRouter team

But the type of model doesn't matter as long as it outputs text. And the throughput doesn't really matter for OpenRouter. It's the users that will care about throughput. The average throughput for mid-sized models on OpenRouter is like 40tps so try to aim for that

[winter jungle]

@rough sundial thanks but I need to contact someone from Openrouter team, do you have any clue how or where to get contact?

[rough sundial]

you just have to wait, since you sent the form already

[steady willow]

I encountered the same issue, please keep me updated on your progress!