Pooled credits
We handle provider access.
$0/ month
Fund once. Route across the catalog.
Fund a wallet →- Provider prices passed through
- Admin-published catalog
- OpenAI-compatible API
- Fallbacks and semantic cache
- Granular spend controls
Usage based
AI gateway pricing
No platform fee; model and provider charges apply.
Bring your own key
Use native provider capacity.
$0/ month
First 1,000,000 BYOK requests each month are free.
Learn how BYOK works →- Encrypted provider-key vault
- Pooled failover when quota runs out
- Per-provider health controls
- Unified usage attribution
- No charge on the first 1M requests
Reseller + enterprise
Govern and rebill organizations.
Custom
Contract terms for advanced governance.
Talk to us →- White-label sub-accounts
- Organization-specific markup rules
- Role and audit controls
- Contracted retention terms
- Dedicated reliability team
Clear math
Buy $1,000 in credits: $1,000 enters your wallet, the fee is $50, and the subtotal is $1,050. Applicable tax is added at checkout.
Ask about pricing →
Cost simulator
Model the month.
See the impact of cache hits and routing policy.
- Cached requests
- —
- Tokens avoided
- —
- Routing effect
- —
Compare
Plan help →The essentials, at a glance.
| Capability | Pooled | BYOK | Enterprise |
|---|---|---|---|
| Monthly platform fee | $0 | $0 | Contract |
| Provider price markup | 0% | Native account | 0% |
| LLMRegistry fee | 5% at funding | 4.5% after 1M requests | Contract |
| Fee basis | Credits purchased | Equivalent provider cost | Usage path |
| Fallback cascades | Included | Included | Custom policies |
| Semantic cache | Project-isolated | Project-isolated | Dedicated option |
| Reseller markup | — | — | Organization controlled |
| Support | Standard | Standard | Dedicated |
Details
Straight answers.
How is model usage billed?+
Pooled input and output tokens are passed through at the resolved provider rate with no inference markup. LLMRegistry charges 5% when credits are funded.
What happens on a cache hit?+
No upstream model request is made. Cache savings appear separately in usage telemetry.
Can I use my own provider keys?+
Yes. The first 1,000,000 BYOK requests per month are free; after that, routing costs 4.5% of equivalent provider cost. Learn how BYOK routing works.
Can I cap spend?+
Yes. Set hard limits by key, project, organization, or billing period, with alerts at 75%, 90%, and 100%.
Ready when you are
Route your first request.
No card required.