TTFT by concurrency
Client time to first streamed token. p50 and p95, warmup excluded.
LLM Inference Benchmark Lab
Reading data/local-cpu-smollm2-135m.json.
Loading summary…
| Field | Value |
|---|
Client time to first streamed token. p50 and p95, warmup excluded.
Successful requests and output tokens, divided by the measurement window.
| Metric | N | Mean | p50 | p95 | p99 |
|---|
Measured requests only. Warmup failures are listed in the methodology panel.
| Request | Level | Prompt | Type | Message |
|---|