Company
About OpenGray
OpenGray is an independent API gateway for frontier language models. We keep the catalogue small, the pricing legible and the billing honest.
What we do
Most teams do not need a directory of four hundred models. They need one endpoint that reliably reaches the best reasoning model available, with billing they can reconcile at the end of the month. That is the whole product.
We resell capacity from frontier labs through a single OpenAI-compatible interface, meter it per token, and charge in credits so there is never a surprise invoice. You hold a balance, you spend it, and you can see exactly where every credit went.
How the platform is built
Model routing
Requests are proxied to frontier providers over OpenRouter, which gives us multi-provider failover without vendor lock-in.
Accounts and metering
Balances, keys and per-request usage live in Postgres on Supabase, with row-level security and atomic credit operations.
Payments
Checkout, cards and receipts are handled by Whop. We never see or store card data.
Credentials
API keys are stored as SHA-256 hashes, passwords as bcrypt digests, and sign-in attempts are rate limited.
Operating principles
- Bill what was used. Charges come from the generation's reported token counts, to six decimal places.
- No lock-in. Pay-as-you-go credits, no contract, no notice period. Stop sending requests and you stop spending.
- Say what the model is. Slugs map to a specific upstream model, published on every model page.
- Answer the phone. Support goes to a person, at the number and address below.