Nova
Gateway Health
LiveRouting reliability, provider performance, and spend across your AI gateway
Requests
515.6K
3.5%Across all projects
Success rate
99.37%
0 ppAfter gateway recovery
P95 TTFT
556ms
2.1%Time to first token
P95 latency
1.65s
0.2%End-to-end response
Fallback rate
2.29%
0.1 pp2.9% retried upstream
Cache savings
$1,412
3.1%34.2% cache hit rate
Primary route
Served by the first selected provider
62.8%
324K
Cache served
Returned without an upstream request
34.2%
176.5K
Provider fallback
Same model, alternate provider
1.7%
8.9K
Model fallback
Backup model completed the request
0.6%
2.8K
Policy blocked
Stopped by gateway policy
0.2%
1.2K
Failed
No route completed successfully
0.4%
2.1K
| Provider | Status | Requests | Route share | Uptime | P95 TTFT | Throughput | Errors |
|---|---|---|---|---|---|---|---|
| OpenAI | operational | 179.4K | 34.8% | 99.99% | 412ms | 121 tok/s | 0.18% |
| Anthropic | degraded | 150.6K | 29.2% | 99.72% | 706ms | 88 tok/s | 1.3% |
| Google Vertex | operational | 110.3K | 21.4% | 99.98% | 338ms | 146 tok/s | 0.24% |
| AWS Bedrock | operational | 50.5K | 9.8% | 99.95% | 462ms | 108 tok/s | 0.31% |
| Mistral | operational | 24.7K | 4.8% | 99.96% | 521ms | 102 tok/s | 0.29% |
$3,651
of $7,500 budget
Elevated rate limiting
Anthropic
Fallback routing is absorbing impact; 96.8% of affected requests recovered.
TTFT returned to baseline
OpenAI
Latency in us-east has remained within SLO for the last 45 minutes.
Routing maintenance completed
Google Vertex
Traffic has returned to the preferred regional endpoint.
99.94% gateway availability
Within the 99.9% monthly SLO
96.8% fallback recovery
Affected requests completed successfully
$3,651 routed spend
Before $1,412 cache savings