I’m a pro user and I’ve opted in to usage-based pricing with a test limit of 10$. However, I’ve gone past my 500 fast quota and now I’m getting queued for next requests, where I’d expect it to use the usage based pricing.
The usage-based pricing is currently only for Claude 3 Opus and a few long-context models. For gpt-4 and claude-3.5-sonnet, you can increase your monthly request limit instead by clicking “change # of fast requests”
Apologies for the confusion here; we intend to transition everything over to the usage-based pricing.