Become a provider

Providers who meet the requirements below can sell inference through RedRouter.

Requirements

An OpenAI-compatible /chat/completions endpoint with streaming, returning token usage on both streaming and non-streaming responses.
A /models endpoint listing each model’s exact ID, pricing in USD, context length, max output, quantization and datacenter locations.
Monthly invoicing for usage.
A published privacy policy and terms stating prompt logging, retention and training use.
Honest model identity: the exact checkpoint and quantization behind each ID, with no silent substitution.

Apply. Tell us about your company and inference service.
Technical review. We review your endpoints, model identity and data policies.
Integration and testing. We check compatibility and usage reporting.
Go live. Approved providers start serving RedRouter customers.

We review applications on a rolling basis and reply by email. Not every application is accepted.

All fields are required.

Company name

Website

Work email

Use your company email. We reply here.

Display name

The name shown to customers on RedRouter (e.g., "Acme AI").

Desired slug

The URL-friendly identifier for your provider (e.g., "acme"). Lowercase letters, numbers, and hyphens only.

Extra details

What makes you a strong provider: hardware, caching, models, volume pricing, rate limits.

URL of your /models endpoint

API base URL

Data policy

Do you train on prompts or completions? If you log prompts, how long do you keep them?

Inference locations

HQ location