RouterPlus pools team demand to win bulk inference token pricing

RouterPlus groups demand from many teams into one large volume deal with a provider, so each team pays the bulk token rate on its share of traffic, according to a post by Rohan Paul (@rohanpaul_ai).

Paul argues this could let small teams get bulk pricing without being large buyers, especially for popular open models.

The service also tests models on a customer's real usage to identify the cheapest model that performs well enough, the post says. The report does not give the provider, the discounted rate, or the volume threshold, and it does not show that the pricing has been achieved in practice.