How we priced Raven: free to start, $20 Pro, and credits that never expire

Raven Pro costs $20 a month, Ultra costs $100, and the API starts at fifty cents per million input tokens with prepaid credits that never expire. Pricing pages have become negotiation openers in this industry, so the numbers are in the first paragraph and the reasoning comes after. This post is the pricing page plus the thinking behind it.

The prices, in full

The Free tier is $0 and it is a real tier, not a trial: the full Raven app, both models as the allowance allows, research, connectors, shared chats, documents. Pro runs $20 a month or $200 a year and carries at least twelve times the Free allowance, with priority processing. Ultra is $100 a month, or $1,000 a year, roughly five times Pro, built for the people who live in the tool. Engineers can add the DCode Coding Plan at $29 a month, on top of any plan or on its own. For the API, credit packs run $5 to $50, pay-as-you-go. A credit you bought in March still spends in December.

Allowances here refill over time rather than counting messages. A long afternoon of heavy reasoning draws more, the tank tops itself back up either way, and the app always shows your percentages, so nothing cuts you off at noon and calls it fair.

Why free is a real tier

Because frontier access should not be means-tested. A student, a bootstrapped founder, and a federal ministry have wildly different budgets and equally real problems. The Free tier is not a crippled demo. It is the full product with a smaller tank, it exists because the mission is access, and "access, for people who can afford it" is a different sentence.

Why Pro costs what the others charge

Twenty dollars is what the industry settled on for consumer AI subscriptions. It is a fair number. We match it on purpose and then try to over-deliver: Pro carries at least twelve times the Free allowance, both models, priority processing, the whole app. Where competitors use the $20 tier to advertise and the $200 tier to deliver, we tried to price the deliverable tier at the familiar price.

Ultra at $100 exists because some people genuinely need several times more capacity, and pretending one size fits all would just get them rate-limited in private. It is the same product with a much bigger tank. After Ultra there is nothing left to upsell you into: no enterprise-only features, and no "contact sales" wall between you and the good models.

The API, to the cent

Raven Flash bills $0.50 per 1M input tokens and $2.50 per 1M output. Raven Max bills $1.25 and $6.25. Cached input drops to $0.05 and $0.13 respectively, applied automatically. The rates are published, the metering is exact, and the credits you buy keep until you use them, because "expires in 90 days" is a small way of saying "we hope you forget." We would rather you remember us for the opposite.

What Founding Members get

For the first cohort of early adopters, annual pricing is locked lower for the entire first year: Pro at $160, Ultra at $800, and the DCode Coding Plan at $230, plus a permanent golden Founding Member badge across every surface. It is our thank-you to the people who funded a bootstrapped lab by showing up early, and like the rest of this page, it is written down plainly: full details on the pricing page.

That is the whole structure. You will not find asterisks doing heavy lifting here, or tiers that exist mainly to make the other tiers look reasonable. A free tier you can build a habit on. A paid tier your accountant will not question, and an API bill you can predict to the cent.