Z.ai: GLM 5.3 Performance
Compare measured TTFT, throughput, uptime, and route health for Z.ai: GLM 5.3 across TrustedRouter providers using metadata-only production probes.
z-ai/glm-5.3
Measured performance
Continuously sampled p50/p95 time-to-first-token (TTFT), effective throughput, and success rate for Z.ai: GLM 5.3. 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 |
|---|---|---|---|---|---|---|
| tinfoil | 752 ms | 3287 ms | 211 tok/s n=1 | 100.00% | 2 provider_error |
54 |
| thinkingmachines | 1280 ms | 3875 ms | 70 tok/s n=2 | 100.00% | — | 10 |
| inceptron | 2482 ms | 6522 ms | 63 tok/s n=1 | 97.62% | — | 42 |
| fireworks | 771 ms | 1239 ms | 128 tok/s n=1 | 100.00% | — | 3 |
| friendli | 843 ms | 843 ms | 298 tok/s n=1 | 100.00% | — | 2 |
| telnyx | 950 ms | 950 ms | 56 tok/s n=1 | 100.00% | — | 1 |
| makora | 1019 ms | 2509 ms | 53 tok/s n=1 | 100.00% | — | 4 |
| engy | 1092 ms | 1283 ms | 140 tok/s n=1 | 100.00% | — | 3 |
| pearl | 1338 ms | 1446 ms | 29 tok/s n=1 | 100.00% | — | 4 |
| sail-research | 1387 ms | 1473 ms | 83 tok/s n=1 | 100.00% | — | 3 |
| crusoe | 1485 ms | 1603 ms | 248 tok/s n=1 | 100.00% | — | 3 |
| akashml | 1563 ms | 2874 ms | 218 tok/s n=1 | 100.00% | — | 4 |
| zro | 1878 ms | 2553 ms | 203 tok/s n=2 | 100.00% | — | 5 |
| wafer | 1893 ms | 3054 ms | 117 tok/s n=1 | 100.00% | — | 4 |
| lyceum | 2072 ms | 2141 ms | 42 tok/s n=1 | 100.00% | — | 2 |
| io-net | 2288 ms | 5132 ms | 180 tok/s n=1 | 100.00% | — | 2 |
| zai | 2728 ms | 5340 ms | 113 tok/s n=1 | 100.00% | — | 7 |
| streamlake | 2732 ms | 3077 ms | 94 tok/s n=1 | 100.00% | — | 2 |
| siliconflow | 3038 ms | 3038 ms | 101 tok/s n=1 | 100.00% | — | 1 |
| zero-g | 3661 ms | 4608 ms | 61 tok/s n=2 | 100.00% | — | 9 |
| nvidia-nim | 4396 ms | 30580 ms | — | 100.00% | — | 2 |
| arcee | 4636 ms | 5742 ms | — | 100.00% | — | 4 |
| cloudflare-workers-ai | — | — | 49 tok/s n=1 | — | — | 0 |
| deepinfra | — | — | 104 tok/s n=1 | — | — | 0 |
| featherless | — | — | 28 tok/s n=1 | — | — | 0 |
| nebius | — | — | 123 tok/s n=1 | — | — | 0 |
| novita | — | — | 85 tok/s n=1 | — | — | 0 |
| parasail | — | — | 191 tok/s n=1 | — | — | 0 |
| privatemode | — | — | — | — | 11 provider_auth_config |
0 |
| tencent | — | — | 71 tok/s n=1 | — | — | 0 |
Full provider & model leaderboard.
33 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.