- Writing speed, one personsingle_user_decode_tps · tok/s
- 13.60 tok/sper person · single · sample count not recorded
- First word afterttft_median_s · s
- 0.81 sper person · median · sample count not recorded
Runs on this setup
Earlier runs keep their own engine versions and outcomes. Workload shapes stay separate.
History and measured sweep
| Date | Shape | Each person · median | Machine total | First word · median | Result | Run |
|---|---|---|---|---|---|---|
2026-10-05 | 1 people | not recorded | not recorded | 0.81 s | Passed | |
2026-10-05 | 1 people | not recorded | not recorded | 8.60 s | Partial | |
2026-10-05 | unknown people | not recorded | not recorded | not recorded | Passed | |
2026-10-05 | 1 people | not recorded | not recorded | not recorded | Not graded | |
2026-10-04 | 1 people | not recorded | not recorded | not recorded | Not graded |
Answer checks
16/16 tasks
Method
Sixteen synthetic exact-output tasks at concurrency 1. Speed and TTFT from separate two-request smoke with thinking disabled; quality from the task cohort.
Limits
Configured 8K context only; no declared customer service targets. Earlier failed smoke remains in the run history.
What ran
Exact identity
- Original publisher model
- Qwen3.8-27B
- Publisher revision
1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0- Variant · format · precision
- INT4 variant · safetensors · INT4
- Weights SHA-256
- not recorded
- Engine · version · feature
- NVIDIA NGC vLLM ·
0.29.0+5013de39.dev· plain - Hardware
- NVIDIA GB10, 128 GB unified memory
- Harness / workload
not recorded- Workload suite · version
not recorded- Harness SHA-256
not recorded- Fixture SHA-256
not recorded
Workload shape
- People in this run
- 1
- Configured window (tokens)
- 8192
- KV cache precision
- bf16
- Feature
- plain
Full reproduction pack available to customers and evaluators on request
All recorded numbers
Metric definitions, scope and statistics
| Metric | Value | Scope | Statistic | Sample count | Basis / source field | Run |
|---|---|---|---|---|---|---|
ttft median s | 0.81 s | per person | median | not recorded | ttft_median_s | |
workload total | 16.00 count | machine total | single | not recorded | workload_total | |
workload passed | 16.00 count | machine total | single | not recorded | workload_passed | |
ttft p95 s | 0.82 s | per person | p95 | not recorded | ttft_p95_s | |
single user decode tps | 13.60 tok/s | per person | single | not recorded | single_user_decode_tps |
Data and citation
Skilak Keep evidence, run c8b7c076e64b38b83c820637, measured 2026-10-05T05:54:15.517300+00:00. https://skilak.ai/keep/resources/evidence/r/c8b7c076e64b38b83c820637
