Happy to clarify, although there’s some important nuance here.
If you’re on the old team plan, there’s $20 of included usage, and that $20 applies across all model usage combined. Once you’ve used up that $20, you switch to on-demand (usage-based) pricing. So in this case, the $20 is a shared amount across everything, and after it’s consumed, additional usage is billed on demand. There is no separate pool for first/third party models.
If you’re on the new teams plan:, his works quite differently, and there’s significantly more usage overall. Instead of a single $20 pool, you have separate usage buckets: a first-party model pool and a third-party API pool.
If you’re on the new pricing, usage of Auto/Composer/Grok draws from the first-party pool first. Once the first-party pool is consumed, additional usage comes from the API pool. This means you can continue using Auto, Composer, and Grok drawing from these buckets as they’re available.