DeepSeek: DeepSeek V4.1 Flash Performance
Try DeepSeek V4.1 Flash on us. Starter credit covers your first requests, then one key reaches all 24 provider routes with automatic failover.
deepseek/deepseek-v4.1-flash
Measured performance
Continuously sampled p50/p95 time-to-first-token (TTFT), effective throughput, and success rate for DeepSeek: DeepSeek V4.1 Flash. Effective throughput uses provider-reported output tokens over complete request time. Unsupported route and probe-configuration rows are separated from provider downtime, and no prompt or output content is stored.
| Provider | p50 TTFT | p95 TTFT | Effective throughput | Uptime | Config excluded | Availability samples |
|---|---|---|---|---|---|---|
| wafer | 879 ms | 3970 ms | — | 99.59% | — | 490 |
| coralbricks | 1268 ms | 3086 ms | 115 tok/s n=1 | 100.00% | — | 22 |
| pearl | 1520 ms | 4874 ms | 87 tok/s n=1 | 100.00% | — | 33 |
| telnyx | 702 ms | 702 ms | — | 100.00% | — | 1 |
| baseten | 886 ms | 1154 ms | — | 100.00% | — | 3 |
| sail-research | 899 ms | 899 ms | — | 100.00% | — | 1 |
| makora | 1001 ms | 1894 ms | — | 100.00% | 2 client_error |
6 |
| lyceum | 1135 ms | 2115 ms | — | 100.00% | — | 6 |
| crusoe | 1137 ms | 2326 ms | 122 tok/s n=1 | 100.00% | 1 provider_error |
3 |
| boundless | 1312 ms | 2400 ms | — | 100.00% | — | 2 |
| engy | 1435 ms | 1482 ms | — | 100.00% | — | 3 |
| zro | 1506 ms | 50965 ms | — | 87.50% | — | 8 |
| atlas-cloud | 1623 ms | 1623 ms | — | 100.00% | — | 1 |
| tinfoil | 2029 ms | 2417 ms | 182 tok/s n=1 | 100.00% | 1 provider_error |
5 |
| siliconflow | 2032 ms | 2032 ms | — | 100.00% | — | 1 |
| featherless | 2913 ms | 2913 ms | — | 100.00% | — | 1 |
| arcee | 3051 ms | 4053 ms | — | 100.00% | — | 4 |
| tencent | 3854 ms | 3854 ms | — | 100.00% | — | 1 |
| nebius | — | — | 103 tok/s n=1 | — | — | 0 |
| together | — | — | — | — | 1 provider_error |
0 |
Full provider & model leaderboard.
24 routes.
More routes give the auto router more room to fail over around provider 429 and 5xx responses.
Gateway overhead is measured separately.
Public status separates TLS/health overhead from full model latency so slow LLMs do not inflate the router metric.
Metadata rollups.
Status samples store latency, outcome, provider, model, route, cost, and region metadata only.
View public status or inspect provider routes.