Z.ai: GLM 5.2 Performance
Compare measured TTFT, throughput, uptime, and route health for Z.ai: GLM 5.2 across TrustedRouter providers using metadata-only production probes.
z-ai/glm-5.2
Measured performance
Continuously sampled p50/p95 time-to-first-token (TTFT), effective throughput, and success rate for Z.ai: GLM 5.2. 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 |
|---|---|---|---|---|---|---|
| together | 594 ms | 2706 ms | 77 tok/s n=1 | 100.00% | — | 2 |
| friendli | 975 ms | 3151 ms | — | 100.00% | — | 7 |
| scaleway | 1410 ms | 4475 ms | — | 100.00% | — | 2 |
| morph | 1525 ms | 2027 ms | 15 tok/s n=1 | 100.00% | — | 3 |
| parasail | 1992 ms | 1992 ms | 138 tok/s n=2 | 100.00% | — | 1 |
| regolo | 2201 ms | 18410 ms | 45 tok/s n=1 | 100.00% | — | 6 |
| engy | 2221 ms | 2426 ms | 47 tok/s n=2 | 100.00% | — | 3 |
| wandb | 2258 ms | 2448 ms | 150 tok/s n=1 | 100.00% | — | 2 |
| gmi | 2509 ms | 4787 ms | — | 100.00% | — | 5 |
| inceptron | 2618 ms | 2822 ms | 58 tok/s n=1 | 100.00% | — | 3 |
| atlas-cloud | 3480 ms | 3480 ms | 84 tok/s n=1 | 100.00% | — | 1 |
| zai | 3538 ms | 5118 ms | 50 tok/s n=2 | 100.00% | — | 5 |
| siliconflow | 4341 ms | 4341 ms | — | 100.00% | — | 1 |
| phala | 4342 ms | 16011 ms | 46 tok/s n=2 | 85.71% | — | 7 |
| arcee | 5607 ms | 5978 ms | — | 100.00% | — | 3 |
| baseten | 11074 ms | 11074 ms | — | 100.00% | — | 8 |
| alibaba | — | — | 56 tok/s n=2 | — | — | 0 |
| deepinfra | — | — | 37 tok/s n=1 | — | — | 0 |
| digitalocean | — | — | 16 tok/s n=2 | — | — | 0 |
| fireworks | — | — | 82 tok/s n=1 | — | — | 0 |
| io-net | — | — | 122 tok/s n=1 | — | — | 0 |
| novita | — | — | 72 tok/s n=1 | — | — | 0 |
| telnyx | — | — | 60 tok/s n=1 | — | — | 0 |
| venice | — | — | 79 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.