MoonshotAI: Kimi K3 Performance
Try Kimi K3 on us. Starter credit covers your first requests, then one key reaches all 28 provider routes with automatic failover.
moonshotai/kimi-k3
Measured performance
Continuously sampled p50/p95 time-to-first-token (TTFT), effective throughput, and success rate for MoonshotAI: Kimi K3. 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 |
|---|---|---|---|---|---|---|
| relace | 1938 ms | 3324 ms | 63 tok/s n=1 | 100.00% | — | 12 |
| akashml | 2809 ms | 6384 ms | 89 tok/s n=2 | 100.00% | — | 339 |
| phala | 5729 ms | 44586 ms | 12 tok/s n=1 | 84.62% | — | 13 |
| together | 21562 ms | 22791 ms | — | 100.00% | — | 67 |
| deepinfra | 789 ms | 789 ms | 37 tok/s n=1 | 100.00% | — | 1 |
| lyceum | 980 ms | 1682 ms | 97 tok/s n=1 | 100.00% | — | 3 |
| fireworks | 1035 ms | 1035 ms | 73 tok/s n=1 | 100.00% | — | 1 |
| featherless | 1191 ms | 1191 ms | 116 tok/s n=1 | 100.00% | — | 1 |
| tinfoil | 1694 ms | 1694 ms | 37 tok/s n=1 | 100.00% | — | 1 |
| baseten | 1717 ms | 2786 ms | 56 tok/s n=1 | 100.00% | — | 3 |
| boundless | 1871 ms | 3524 ms | 103 tok/s n=1 | 100.00% | — | 3 |
| sail-research | 1887 ms | 4182 ms | 197 tok/s n=1 | 100.00% | — | 2 |
| makora | 2100 ms | 2898 ms | 65 tok/s n=2 | 100.00% | — | 5 |
| siliconflow | 2118 ms | 2118 ms | 38 tok/s n=1 | 100.00% | — | 1 |
| telnyx | 2150 ms | 2150 ms | 72 tok/s n=1 | 100.00% | — | 1 |
| venice | 2153 ms | 3033 ms | 110 tok/s n=1 | 100.00% | — | 2 |
| wafer | 2653 ms | 2653 ms | — | 100.00% | — | 1 |
| gmi | 3016 ms | 7120 ms | 106 tok/s n=1 | 100.00% | — | 7 |
| kimi | 3462 ms | 5910 ms | 43 tok/s n=2 | 100.00% | — | 9 |
| cloudflare-workers-ai | 3607 ms | 6460 ms | 24 tok/s n=1 | 100.00% | — | 2 |
| tencent | 4140 ms | 4140 ms | 39 tok/s n=1 | 100.00% | — | 1 |
| novita | 4361 ms | 4361 ms | 23 tok/s n=1 | 100.00% | — | 1 |
| atlas-cloud | 4681 ms | 4681 ms | 41 tok/s n=1 | 100.00% | — | 1 |
| arcee | 5322 ms | 5649 ms | 32 tok/s n=1 | 100.00% | — | 3 |
| chutes | — | — | 21 tok/s n=1 | — | — | 0 |
| engy | — | — | — | — | 9 provider_error |
0 |
| io-net | — | — | 120 tok/s n=1 | — | — | 0 |
| nebius | — | — | 127 tok/s n=2 | — | — | 0 |
Full provider & model leaderboard.
28 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.