DeepSeek: DeepSeek V4 Flash 0731 Performance
Try DeepSeek V4 Flash 0731 on us. Starter credit covers your first requests, then one key reaches all 26 provider routes with automatic failover.
deepseek/deepseek-v4-flash-0731
Measured performance
Continuously sampled p50/p95 time-to-first-token (TTFT), effective throughput, and success rate for DeepSeek: DeepSeek V4 Flash 0731. 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 |
|---|---|---|---|---|---|---|
| baseten | 1553 ms | 2278 ms | — | 100.00% | 1 provider_error |
70 |
| relace | 1691 ms | 2757 ms | — | 100.00% | — | 11 |
| confidential-ai | 1907 ms | 2834 ms | 97 tok/s n=1 | 96.55% | — | 29 |
| inceptron | 2732 ms | 66732 ms | — | 93.33% | 1 insufficient_quota |
15 |
| engy | 865 ms | 904 ms | — | 100.00% | — | 2 |
| lyceum | 868 ms | 868 ms | — | 100.00% | — | 1 |
| boundless | 912 ms | 912 ms | — | 100.00% | — | 1 |
| venice | 1064 ms | 1064 ms | — | 100.00% | — | 1 |
| sail-research | 1239 ms | 3219 ms | — | 100.00% | — | 4 |
| mancer | 1280 ms | 2150 ms | — | 100.00% | — | 6 |
| telnyx | 1299 ms | 1299 ms | — | 100.00% | 1 provider_error |
1 |
| io-net | 1794 ms | 1794 ms | — | 100.00% | — | 1 |
| streamlake | 1894 ms | 1894 ms | — | 100.00% | — | 1 |
| siliconflow | 1955 ms | 1955 ms | — | 100.00% | 1 provider_error |
1 |
| byteplus | 2179 ms | 2609 ms | — | 100.00% | — | 5 |
| wandb | 2453 ms | 2709 ms | — | 100.00% | — | 3 |
| cloudflare-workers-ai | 2670 ms | 3766 ms | — | 100.00% | — | 2 |
| nebius | 2793 ms | 2793 ms | — | 100.00% | 1 provider_error |
1 |
| nextbit | 3247 ms | 4625 ms | 133 tok/s n=1 | 100.00% | — | 3 |
| deepinfra | — | — | — | 100.00% | — | 172 |
Full provider & model leaderboard.
26 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.