Pricing

One layer. Every model.

One subscription for the intelligence layer. Your provider bills you separately, at their price — we never take a cut of it. Lower-cost regions receive automatically lower pricing at checkout.

Free

Try the layer before you bring a key.

$0/mo

1M tokens / mo · 3 API keys

  • All 5 frontier models
  • Persistent memory & knowledge objects
  • 3 API keys
  • No overage — hard stop at quota
Start free

Starter

Wire memory into one real application.

$9/mo

10M tokens / mo · Unlimited keys

  • Everything in Free
  • Unlimited API keys
  • Briefings — standing instructions per key
  • Multi-provider routing
  • Overage at $1.50 per extra 1M tokens
Choose Starter

Pro

Production traffic, with headroom per key.

$29/mo

60M tokens / mo · Unlimited keys

  • Everything in Starter
  • Knowledge sets — files a key consults on every call
  • Per-key spend ceilings
  • Overage at $0.80 per extra 1M tokens
Choose Pro

Scale

Teams and a fleet of keys.

$99/mo

250M tokens / mo · Unlimited keys

  • Everything in Pro
  • Team seats
  • SSO with your own IdP
  • Overage at $0.50 per extra 1M tokens
Choose Scale

Compare tiers

FeatureFreeStarterProScale
Tokens counted / month1M10M60M250M
API keys3UnlimitedUnlimitedUnlimited
All 5 frontier models
Persistent memory & knowledge objects
Briefings (standing instructions)
Multi-provider routing
Knowledge sets
Per-key spend ceilings
Team seats
SSO with your own IdP
Overage / extra 1M tokensHard stop$1.50$0.80$0.50

What “tokens counted” means

The quota above counts the tokens your own provider key reports back to us — the same tokens on your provider invoice. Retrieving and assembling context adds tokens to every request; we tell you exactly how many we added, per message and in aggregate. We don't publish what we added or how it was structured — the structure is ours.

Frequently asked

What am I paying for if I bring my own key?

Your provider bills you for tokens, at their price, with no markup from us. The subscription pays for the layer wrapped around those tokens: memory that persists across sessions, knowledge objects distilled from your conversations, semantic retrieval over everything a key has seen, and the ability to move a key between all five models without losing any of it.

What happens when I hit my token quota?

On Free, calls stop until the quota resets — no surprise bill. On every paid tier, overage kicks in automatically at a fixed per-token rate that falls as you move up tiers, priced so upgrading is always cheaper than running heavy overage. You can check usage against quota at any time.

Do you see my provider key?

No. Your key is sealed in a secrets vault the moment you paste it in. We hold a reference to it, never the value — it's validated once against your provider, then write-only from that point forward. Nobody, including our own admins, can read it back through the product.

Can I cancel?

Yes, at any time, from your billing settings. Your plan stays active through the end of the period you already paid for, and your keys keep working until then. After that they stop — nothing is deleted out from under you mid-cycle.