typed++ is live: the deepest reasoning we can buy

The short version

typed++ is live. It is a new model family that sits above typed-max at the top of our lineup, and it is the deepest reasoning we can currently buy.

It is not a new plan. There is no upgrade to purchase, no SKU, no separate quota to manage. If you have a typed subscription today -- any tier, including Starter -- you can use typed++ right now by naming it:

typed --model typed-plus-plus

The one thing to understand before you turn it on: it draws from your existing monthly budget about 10x faster. That is the whole trade. The rest of this post explains why, because we would rather you understand the economics than discover them at the end of the month.

Why it is a separate family, not another notch on the ladder

The typed-* tiers -- typed-max through typed-low -- are one model family at five reasoning depths. Moving from typed-medium to typed-max buys you more thinking on the same underlying capability.

typed++ is a different family entirely. It solves a different problem: not "think longer about this," but "be more capable at this." When you have a problem where the ceiling is capability rather than reasoning headroom -- gnarly architecture calls, ambiguous debugging, the class of task where a better model wins and a longer think does not -- that is what typed++ is for.

Bench data from our own tier work makes the distinction concrete: on the cheaper tiers, raising the reasoning ceiling did not move results, because capability was the bottleneck, not headroom. Raising the ceiling only pays when the model underneath can use it. typed++ raises the model.

The honest part: what it costs you

typed's quota has always been cost-based, not token-based. Each plan carries a fixed ceiling on what we spend serving you, and every request debits its real cost against that ceiling.

typed++ is roughly 10x the cost to run as our default family. Because the budget is denominated in cost rather than tokens, that means:

• Your monthly ceiling is unchanged.
• A typed++ request debits about 10x what the same request costs on the default family.
• So at the extreme, an all-typed++ month gets you roughly one tenth the tokens.

We are not marking it up. Our margin on a typed++ request is identical to a default-family request -- the same ceiling, the same percentage -- and that equality is locked by a test in our own build so it cannot quietly drift. You are paying the real cost difference, not a premium for the label.

This is also why typed++ is a toggle instead of a plan. A separate priced tier would have meant a new SKU, a new quota envelope, and a decision from you at signup about a model you had not tried yet. A toggle means you spend your existing budget where it is worth spending -- typed++ on the hard hour, the default family on the other seven.

One effort level, honestly labeled

At launch typed++ runs at its top effort level only. That is the only level the family currently exposes. Additional levels are promised upstream, and when they land we will expose them as real, separately-named tiers.

What we will not do is simulate them. We could ship typed-plus-plus-high tomorrow by capping a token budget and calling it a lower tier -- it would look like a fuller product page. It would also be a lie about what you are buying. When those levels are real, they get names. Until then, there is one.

When to reach for it

Reach for typed++ when the problem is hard in a way that a bigger thinking budget will not fix:

• Architecture and design calls where the tradeoffs are genuinely unclear
• Debugging where the symptom and the cause are far apart
• Work where being 90% right is the same as being wrong

Stay on typed-xhigh -- still the production default -- for everything else. It is fast, coder-tuned, cache-reliable, and it is the right tool for the large majority of real work. Burning 10x budget on a routine refactor is not thoroughness, it is waste.

How to turn it on

Per invocation:

typed --model typed-plus-plus

Or via the API, as the model field:

"model": "typed-plus-plus"

It appears at the top of /v1/models as Typed Plus Plus. No header, no environment variable, no flag to enable first -- naming the model is what selects it.

If you want to see what it is doing to your budget, the dashboard's usage page breaks spend down by request, and we return quota headers on every response.