The Qwen3.5 native vision-language Flash models are built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. Compared to the...
Share of successful requests per carrier, measured over the last 5 minutes, 30 minutes and day. When one carrier dips, the router fails over to the next.
| Provider | Status | Last 5m | Last 30m | Last day |
|---|---|---|---|---|
| Operational | 100.00% | 100.00% | 100.00% |