Credits are the unit. Routed usage, intelligence features, and plan allotments all bill in them, and a credit is worth the same everywhere. Start free, pick Pro, Max or Ultra (each step up raises your limits and makes models cheaper), or scale on Enterprise. Routed tokens stay at provider cost, 0% on top.
500 credits a month once you verify a card (a one-time $1 charge, added to your balance). Route on credits at each provider’s list price with nothing on top.
The managed layer around the gateway: identity, policy, data residency, and the intelligence engagement, scoped with your team at the lowest per-credit rate.
Book a call| Free | Pro · Max · Ultra | Enterprise | |
|---|---|---|---|
| Plans and credits | |||
| Included credits per month | 500 (after $1 card verification) | 2,000 to 1,000,000 | Custom |
| Free models (Pro and up; Ultra adds more) | Not included | DeepSeek V4.1 Flash, GPT-6 Luna, Jev | Custom |
| Discounts on other models (Pro / Max / Ultra) | Up to 25% | Up to 50% / 75% / 45% | Custom |
| Rate limits | Tier 1 | Tier 2 / 3 / 4 | Tier 5 and custom |
| Early releases | Not included | Max and Ultra | Included |
| Flat 1¢ per credit, 0% token markup | Included | Included | Included |
| Buy more credits anytime (top-ups) | Included | Included | Included |
| Markup on routed tokens | 0% | 0% | 0% |
| One bill across providers | Included | Included | Included |
| Gateway | |||
| OpenAI-compatible /v1 endpoint | Included | Included | Included |
| Every hosted provider | Included | Included | Included |
| Your own provider keys (BYOK) | Not included | Included | Included |
| Local and custom models | Not included | Included | Included |
| Provider failover | Included | Included | Included |
| Controls | |||
| Budgets and hard caps | Included | Included | Included |
| API key management | Included | Included | Included |
| Model allowlists per key or agent | Included | Included | Included |
| Dashboard | Included | Included | Included |
| Observability | |||
| Usage by agent, person, model, day | Included | Included | Included |
| Request logs with route and cost | Included | Included | Included |
| Live traffic view | Included | Included | Included |
| Security | |||
| Route only to ZDR providers | Included | Included | Included |
| No-training provider policy | Included | Included | Included |
| Provider allowlists | Included | Included | Included |
| Provider keys never reach agents | Included | Included | Included |
| Intelligence | |||
| Per-prompt model optimization | Not included | Included | Included |
| Caching | Not included | Included | Included |
| Intelligence features, billed in credits | Not included | Included | Included |
| A model you own, trained on your traffic | Not included | Not included | Included |
| Support | |||
| Community (Discord and GitHub) | Included | Included | Included |
| Dedicated support | Not included | Included | Included |
| Slack support | Not included | Not included | Included |
| Security reviews | Not included | Not included | Included |
Routing policies, budgets, and allowlists are enforced at the gateway on every request. No code changes.
The posture an enterprise rollout asks for: what ships today, what we scope with your team, and what is on the roadmap. For the intelligence layer, security reviews, and dedicated support, talk to us.
Book a callSeparate orgs, keys, and billing under one account.
OAuth sign-in for the whole team, no passwords required.
Usage, spend, requests, TTFT, and token counts at key and team scope.
Spend by agent, person, model, or day, pulled into your own systems.
Create, view, and delete keys from the dashboard or the API.
Route only to providers under a ZDR agreement, enforced in policy.
Route only to providers that will not train on customer data.
Failover across providers and pooled accounts on capacity errors.
Every request recorded with its route, tokens, and cost.
Hard caps and model allowlists, enforced on every request.
SOC 1 compliance, report available on request.
Enterprise identity providers and directory sync.
Expiring tokens in place of static keys.
Fine-grained roles beyond the built-in org roles.
One policy set applied across every team and key.
Changes reviewed before they land, with change history.
Customer-managed keys for stored provider credentials.
Private connectivity between your VPC and the gateway.
Regional routing and storage constraints.
A named contact and a shared channel.
For self-hosted deployments we operate with you.
Dedicated capacity for the intelligence layer.
Spend projections and threshold notifications.
A credit is one standardized unit of usage. Routed tokens spend credits at the provider's list price, and the intelligence features spend them too: one unit across everything, worth the same wherever it is spent. Plans and top-ups are just how you buy credits.
Yes. A routed token spends credits at the provider's list price. We add $0.00 per token, on every plan. What you pay for is the credits themselves: the plan you pick and the intelligence layer, never a percentage of your traffic.
A plan is a monthly credit allotment at a flat 1¢ per credit, the same rate as a top-up. Top-ups let anyone buy credits without a plan. What a plan adds is the unlocks: BYOK and local models, actionable intelligence, prompt storage off, zero data retention, higher rate limits, and discounts on models. $20 to $199 a month is Pro, $200 to $1,999 is Max, $2,000 and up is Ultra; each step up raises your limits and deepens the discounts, and the discounts cover your plan's monthly credits of usage.
Requests on credits stop at your caps, so nothing silently overruns. Add more from the dashboard, and anything routed through your own provider keys keeps working the whole time.
Yes, on Pro, Max, Ultra and Enterprise. The gateway routes through your keys first, so existing provider commitments and discounts flow through untouched, and traffic on your own keys is never rate limited by us.
The whole gateway: the OpenAI-compatible /v1 endpoint, key management, budgets, provider waterfalls, and the usage API. uvx --from experiential exp run starts it on your own infrastructure.
One base URL swap. Provider prices, hard budgets, every token attributed.