Provider tiers
Every model is served by economy, standard and premium providers. Requests are routed cheapest first with automatic fallback, so you get up to 80% off the official rate and the same response whichever tier answers.
Routing
A request is never pinned to one upstream. It moves through three layers of fallback in milliseconds, and settles on the cheapest healthy provider that can serve the model.
If a request fails, equivalent models are tried next — Claude to GPT to DeepSeek.
Providers are tried cheapest first: economy, then standard, then premium.
Unhealthy providers are skipped automatically instead of retried.
If a cheaper tier is unavailable, the request continues to the next one and is billed at that tier's rate.
Tier details
Each tier is a pool of upstream providers serving the identical model catalog. Pick where price matters and where official SLAs matter; the router does the rest.
The lowest-cost pool of providers serving the same models. Tried first, so most requests settle at the cheapest rate.
High-volume and cost-sensitive workloads where price matters more than shaving off the last few milliseconds.
Balanced reliability and cost. A mix of official APIs and trusted aggregators. Tried after economy.
Everyday production workloads. Same model quality as premium, at a lower price.
Official APIs from OpenAI, Anthropic and Google. Highest uptime, fastest response, full SLA support.
Mission-critical workloads needing official SLAs or features exclusive to direct provider APIs.
Billing
No monthly commitment. You top up a balance and each request is charged at the rate of the tier that served it. Which balance pays never changes the price — only which tiers it may pay for.
Credits you purchase. Pays on every tier, including premium.
Referral rewards and gift codes. Pays on economy and standard only.
FAQ
Yes. You can configure tier preferences per model in your dashboard. Tiers are tried cheapest first — economy, standard, premium — but you can customize this. For example, enable premium only if you need official APIs exclusively.
No. All three tiers serve the exact same models. A request to the same model returns identical responses whether routed through economy, standard or premium providers. The only differences are cost, latency and reliability.
The routing system automatically falls back to premium so your request still succeeds — charged at 1.00x instead of 0.50x. Circuit breakers prevent repeated attempts to unhealthy providers.
Requests are rejected with a 402 Payment Required response until you top up. Premium requests also fail if you only hold bonus balance, since bonus credits cannot pay for premium providers.
Top up once. Every request is routed to the cheapest healthy provider automatically.