Groq
p50 96ms · p99 230ms
operational
One endpoint in front of six LLM providers. Every app calls the gateway; the gateway picks the fastest healthy provider, retries, and fails over automatically when one goes down. This console is what the operator sees.
Requests cascade through this chain. Kill one with “Simulate outage” and watch traffic re-route without a single dropped request.
Green: primary routes. Cyan: re-routed traffic during a failover.
Every call from every app in the fleet, as it happens. New rows stream in live.
| time | route | model | provider | tokens | latency | status |
|---|---|---|---|---|---|---|
| 14:41:59.87 | /v1/chat | qwen-3-32b | Groq | 1,377 | 118ms | 200 |
| 14:41:55.29 | /v1/embeddings | mixtral-9x22 | Gemini | 1,946 | 357ms | 200 |
| 14:41:51.28 | /v1/chat | qwen-3-32b | Cerebras | 2,581 | 91ms | 200 |
| 14:40:47.51 | /v1/audio | qwen-3-32b | OpenRouter | 2,114 | 412ms | 200 |
| 14:40:43.54 | /v1/chat | llama-4-70b | OpenRouter | 3,618 | 335ms | 200 |
| 14:40:39.85 | /v1/chat | llama-4-70b | Cloudflare | 1,939 | 239ms | 200 |
| 14:39:35.31 | /v1/chat | qwen-3-32b | OpenRouter | 2,083 | 396ms | 200 |
| 14:39:31.82 | /v1/embeddings | glm-5-air | Cerebras | 2,291 | 90ms | 200 |
| 14:39:27.24 | /v1/embeddings | glm-5-air | Cerebras | 2,180 | 99ms | 200 |
| 14:38:23.22 | /v1/chat | glm-5-air | Gemini | 1,377 | 391ms | 200 |
| 14:38:19.63 | /v1/chat | mixtral-9x22 | Cerebras | 1,234 | 117ms | 200 |
| 14:38:15.23 | /v1/embeddings | mixtral-9x22 | Cloudflare | 2,531 | 230ms | 200 |
| 14:37:11.72 | /v1/chat | glm-5-air | Cloudflare | 1,245 | 207ms | 200 |
| 14:37:07.37 | /v1/chat | qwen-3-32b | Gemini | 493 | 557ms | 200 |