One membership.
More ways to create.

A simpler way to work across AI models, without juggling a new conversation for every choice. That is the membership we are building.

Everyday AI, premium AI and API access

Our proposed plans separate generous everyday interactive use, an allowance for more expensive models, and explicitly metered developer or automated workloads. Final pricing, eligible models and usage limits will be published before purchase becomes possible.

When a premium allowance is used, an eligible lower-cost mode could continue in the same conversation only when separately funded capacity remains. The active model and its allowance should remain visible.

What a million tokens actually means in a plan

A token allowance needs a model, direction and charging rule to be meaningful. A plan must explain how input, output, cached input, reasoning and other charged units count where applicable. It must also distinguish one request’s maximum size from the total monthly allowance.

We have not yet contracted or validated an included token volume. This preview therefore does not advertise millions of included tokens, a completed model catalogue or unlimited usage. A headline capacity number will be published only when supported by supplier terms and tested costs.

Continuity without hidden substitutions

Our planned Auto mode would favour an available, permitted model that fits the task and budget. Model lock would retain the chosen model. Customers would see the responder’s identity rather than receiving an undisclosed substitution.

Before memberships open

We need validated commercial supply, measured request costs, secure customer accounts, durable conversation storage and subscription billing that activates and revokes access correctly. The current preview does not take payment or provision API access.