Enterprise-grade commitments on uptime availability, latency targets, upstream reliability, and financial credit remedies for global API developers and partners.
This Service Level Agreement ("SLA") governs the availability, latency, security, and quality of AI Model Gateway routing endpoints provided by HỘ KINH DOANH NGUYỄN GIA HUY (including OpenAI-compatible /v1/chat/completions, /v1/models, /v1/embeddings, and Anthropic /v1/messages) to enterprise accounts, resellers, and international client applications.
We guarantee a Monthly Uptime Percentage of at least 99.9% across all active gateway ingress clusters. The availability percentage is calculated on a per-calendar-month basis:
If the gateway fails to maintain the guaranteed monthly uptime, customers are eligible for Service Credits applied directly to their account balance as follows:
| Monthly Uptime Percentage | Service Credit Percentage | Issuance Method |
|---|---|---|
| ≥ 99.9% | Normal SLA compliant | — |
| 99.0% – 99.89% | 10% of monthly bill | Credited to wallet balance |
| 95.0% – 98.99% | 25% of monthly bill | Credited to wallet balance |
| < 95.0% | 50% of monthly bill | Credited to wallet balance |
Our gateway operates an intelligent multi-region proxy architecture integrated with verified direct upstream providers, including Google Cloud Vertex AI, Anthropic PBC, OpenAI LLC, Meta Llama Clusters, and DeepSeek Platform.
Our gateway maintains strict Time-to-First-Token (TTFT) benchmarks across flagship models under normal operating loads:
Our infrastructure engineering team operates 24/7/365 active monitoring with strict contractual response times by incident severity:
| Severity Level | Description | Initial Response Time | Support Channel |
|---|---|---|---|
| Severity 1 (Critical) | Total gateway outage affecting all requests across clusters. | < 15 phút (24/7) | Hotline / Telegram VIP / Ticket |
| Severity 2 (Major) | Degraded performance or high error rates on a specific model endpoint. | < 1 giờ | Telegram / Email Support |
| Severity 3 (Standard) | General technical inquiries, billing queries, commercial invoice requests. | < 4 giờ | Live Chat / Email Ticket |
Downtime excludes: (a) Scheduled maintenance announced at least 48 hours in advance (typically 02:00 - 05:00 UTC+7); (b) Client-side application errors, bad request payload, or self-inflicted quota depletion; (c) Force majeure events including international submarine cable outages.
To claim Service Credits, submit a notice within 30 days of the incident month to [email protected] with request logs or timestamp evidence. Approved credits are deposited into your wallet balance within 3 business days.
068208000472