Everything on the self-serve platform, plus the billing and support arrangements production teams actually need: invoice and wire transfer, committed-spend terms, custom routing, and a human who answers.
Talk to us Start self-servePay by USD wire against an invoice instead of card or USDT. Suitable for companies that cannot use consumer payment rails.
The per-token rate stays at provider list price with zero platform markup; for sustained volume we can layer a committed-spend agreement on top.
Pin specific upstream providers, set fallback order between equivalent models, and route by region or latency requirement.
A named contact for production issues, with response-time commitments available in the agreement.
Data-handling summary, billing provenance feed, request-log export, and a walkthrough of how calls route upstream.
Moving from a direct provider account or another gateway: we map model ids, verify parity on your prompts, and stage the cutover.
| Uptime | Measured, not aspirational — rolling 30-day figure and every probe are public at status.json (currently 99.87% over 2,332 probes, 0 incidents in 30 days). |
| Platform fee | 0% — provider list price passed through, multiplier 1.0. Machine-readable at transparency.json. |
| Billing provenance | Per-call cost breakdown published at provenance.json. |
| Rate limits | 60 requests/min on paid accounts; higher concurrency available on request. |
| Models | 111 live models across DeepSeek, Qwen, GLM, Kimi, Doubao/Seedance, MiniMax, Hunyuan, Kling, Vidu, Tripo. Live list: /v1/models. |
Yes. Teams can be billed by invoice with bank transfer (USD wire) instead of card or USDT. Contact us with your company name and expected monthly spend and we will set it up.
Pricing passes through the upstream provider rate with zero platform markup. For sustained volume we can discuss committed-spend terms on top of that pass-through price; the per-token rate itself is already the provider list price.
We publish measured uptime rather than an aspirational number: the rolling 30-day figure with every probe is at status.json, and the incident history is public. Enterprise agreements can add response-time commitments for support tickets.
Yes. We can provide our data-handling summary, the billing provenance feed, and a walkthrough of how requests route to upstream providers. Request logs can be exported from the dashboard on request.
Email [email protected] with your company name, expected monthly volume, and which models you need. We reply within one business day.