Sample data. Not a customer bill.
Provider health
PlannedUptime, errors, latency, fallbacks and gateway overhead.
View
Gateway overhead p50
20 ms
Sample gateway overhead. The gateway is not deployed.
Gateway overhead p95
24 ms
Fallback events
2
Recovered requests, HTTP 200
| Provider | Requests | Error rate | Success without fallback | p50 | p95 | Overhead p50 |
|---|---|---|---|---|---|---|
| OpenAI | 12 | 16.7% | 83.33% | 675 ms | 765 ms | 20 ms |
| Anthropic | 0 | 0.0% | No samples | — | — | — |
| Google Gemini | 0 | 0.0% | No samples | — | — | — |
| Mistral | 0 | 0.0% | No samples | — | — | — |
| DeepSeek | 0 | 0.0% | No samples | — | — | — |
| Qwen | 0 | 0.0% | No samples | — | — | — |
| Cohere | 0 | 0.0% | No samples | — | — | — |
| Meta hosts | 0 | 0.0% | No samples | — | — | — |
| Self-hosted | 0 | 0.0% | No samples | — | — | — |
Fallback timeline
Open in logs- req_harbour_0001harbour-replyopenai600 ms
- req_harbour_0012harbour-replyopenai765 ms