LLM provider status
Official status from each provider and the uptime LLM Gateway measured on real traffic over the last 24 hours. When a provider fails, LLM Gateway fails over to another one serving the same model.
1 provider reporting issues
| Provider | Official status | Measured uptime (24h) | Details |
|---|---|---|---|
| Cloudflare | Major outage | 100% | |
| Anthropic | Operational | 99.50% | |
| Cerebras | Operational | — | |
| ElevenLabs | Operational | — | |
| Fireworks AI | Operational | 99.99% | |
| Groq | Operational | — | |
| MiniMax | Operational | 99.97% | |
| Mistral AI | Operational | — | |
| Moonshot AI | Operational | 100% | |
| Nebius AI | Operational | — | |
| OpenAI | Operational | 99.97% | |
| Perplexity | Operational | — | |
| Alibaba Cloud | No official feed | 99.05% | |
| AtlasCloud | No official feed | — | |
| AWS Bedrock | No official feed | 98.97% | |
| AWS Mantle | No official feed | — | |
| Azure | No official feed | 99.89% | |
| Azure Anthropic | No official feed | — | |
| Baidu | No official feed | — | |
| ByteDance | No official feed | — | |
| CanopyWave | No official feed | — | |
| DeepInfra | No official feed | 99.97% | |
| DeepSeek | No official feed | 99.97% | |
| fal | No official feed | — | |
| Google AI Studio | No official feed | 99.65% | |
| Google Vertex AI | No official feed | 96.25% | |
| Inference.net | No official feed | 96.45% | |
| Meta | No official feed | 99.66% | |
| Meta Contributor | No official feed | 99.89% | |
| NovitaAI | No official feed | 99.99% | |
| RanoAI | No official feed | — | |
| Runpod | No official feed | — | |
| Sakana AI | No official feed | — | |
| Tencent Cloud | No official feed | — | |
| Together AI | No official feed | 100% | |
| TypeSafe AI | No official feed | 99.99% | |
| Vertex AI (Anthropic) | No official feed | — | |
| Vertex AI (OpenAI-compatible) | No official feed | — | |
| xAI | No official feed | — | |
| Xiaomi | No official feed | 100% | |
| Z AI | No official feed | 99.64% |
Measured uptime counts requests on LLM Gateway credits that ended without a provider or gateway error; invalid requests and customers' own provider keys are excluded. Updated every five minutes.