- First word afterttft_median_s · s
- 1,437.85 sper person · median of 4
- Slowest first wordttft_max_s · s
- 2,300.67 sper person · max of 4
- Machine totalaggregate_decode_tps · tok/s
- 0.24 tok/smachine total · single · sample count not recorded
- Machine total including first-word waitaggregate_e2e_tps · tok/s
- 0.18 tok/smachine total · single · sample count not recorded
- Everyone answered inwall_seconds · s
- 2,309.11 smachine total · single · sample count not recorded
- Peak memory, whole hostpeak_host_unavailable_gib · GiB
- 82.89 GiBmachine total · max · 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 |
|---|---|---|---|---|---|---|
2026-10-06 | 5 people · shared · warm | not recorded | 15.19 tok/s | 3.39 s | Passed | |
2026-10-06 | 5 people · shared · cold | not recorded | 0.23 tok/s | 1,729.64 s | Failed | |
2026-10-06 | 5 people · distinct · cold | not recorded | 1.31 tok/s | 306.97 s | Failed | |
2026-10-06 | 5 people · distinct · cold | not recorded | 0.57 tok/s | 700.47 s | Failed | |
2026-10-06 | 5 people · distinct · warm | not recorded | 13.98 tok/s | 3.42 s | Passed | |
2026-10-06 | 5 people · distinct · cold | not recorded | 0.23 tok/s | 1,733.38 s | Failed | |
2026-10-06 | 4 people · distinct · cold | not recorded | 0.24 tok/s | 1,437.85 s | Failed | |
2026-10-06 | 3 people · distinct · cold | not recorded | 0.28 tok/s | 1,147.20 s | Failed | |
2026-10-06 | 2 people · distinct · cold | not recorded | 0.37 tok/s | 857.60 s | Failed | |
2026-10-06 | 1 people · distinct · cold | not recorded | 13.02 tok/s | 562.34 s | Passed |
Each person in this run
First-word wait against writing speed after that word, including pauses. Earlier streams can pause while other people’s prompts are read. The table preserves filled input, cached-token counts and each person’s checks.
Per-person table · all measured request rows
| Person | Filled input | Cached tokens | Output | First word | Writing after first word | Total time | Checks | Run |
|---|---|---|---|---|---|---|---|---|
1 | 250,011 | 0 | 99 | 568.80 s | 1.02 tok/sone person · includes waits | 664.50 s | 3/3 Hidden facts found · completed · checks passed | |
3 | 250,011 | 0 | 109 | 1,146.74 s | 0.96 tok/sone person · includes waits | 1,258.96 s | 3/3 Hidden facts found · completed · checks passed | |
4 | 250,011 | 0 | 103 | 1,728.95 s | 0.97 tok/sone person · includes waits | 1,833.82 s | 3/3 Hidden facts found · completed · checks passed | |
2 | 250,011 | 0 | 108 | 2,300.67 s | 12.68 tok/sone person · includes waits | 2,309.11 s | 3/3 Hidden facts found · completed · checks passed |
Per-person table · all measured request rows
| Person | Filled input | Cached tokens | Output | First word | Writing after first word | Total time | Checks | Run |
|---|---|---|---|---|---|---|---|---|
1 | 250,011 | 0 | 99 | 568.80 s | 1.02 tok/sone person · includes waits | 664.50 s | 3/3 Hidden facts found · completed · checks passed | |
3 | 250,011 | 0 | 109 | 1,146.74 s | 0.96 tok/sone person · includes waits | 1,258.96 s | 3/3 Hidden facts found · completed · checks passed | |
4 | 250,011 | 0 | 103 | 1,728.95 s | 0.97 tok/sone person · includes waits | 1,833.82 s | 3/3 Hidden facts found · completed · checks passed | |
2 | 250,011 | 0 | 108 | 2,300.67 s | 12.68 tok/sone person · includes waits | 2,309.11 s | 3/3 Hidden facts found · completed · checks passed |
Answer checks
12/12 Hidden facts found4/4 request checks
Method
Synchronized long public-domain prompts, three exact hidden facts per user. Decode rate uses completion tokens minus one over first output through stream completion; effective prefill includes scheduling and first-token latency. Aggregate decode uses the union window from first output to final completion; end-to-end rate includes prefill.
Limits
Lab observation, not customer acceptance. Prompt content length excludes the server chat template; actual per-request usage is recorded. Synthesis is requested and retained but semantic accuracy is unassessed. All requested users must overlap in observed decode to pass.
What did not work
- Observed decode overlap was 1 of 4 requested users.
What ran
Exact identity
- Original publisher model
- Qwen3.8-27B
- Publisher revision
1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0- Variant · format · precision
- IQ3_S GGUF variant · GGUF · IQ3_S
- Weights SHA-256
- not recorded
- Engine · version · feature
- llama.cpp CUDA ·
version not recorded· with multi-token prediction - Hardware
- NVIDIA GB10, 128 GB unified memory
- Harness / workload
keep-long-context-concurrency-v1 · 1- Workload suite · version
not recorded- Harness SHA-256
not recorded- Fixture SHA-256
1e01cbcbd88ac9878b889cba44f56c74f5ab023f35e2d28a3bbe343923d909e0
Workload shape
- cache state
- cold
- People in this run
- 4
- Configured window (tokens)
- 262144
- Filled content target (tokens)
- 250000
- fixture variant
- distinct-v1
- KV cache precision
- q8_0
- Output token ceiling
- 512
- scenario
- distinct
- Thinking
- off
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 |
|---|---|---|---|---|---|---|
wall seconds | 2,309.11 s | machine total | single | not recorded | wall_seconds | |
ttft median s | 1,437.85 s | per person | median | 4 | ttft_median_s | |
hidden facts total | 12.00 count | machine total | single | not recorded | needle_total | |
ttft max s | 2,300.67 s | per person | max | 4 | ttft_max_s | |
requests ok | 4.00 count | machine total | single | not recorded | requests_ok | |
hidden facts correct | 12.00 count | machine total | single | not recorded | needle_correct | |
decode overlap | 1.00 count | machine total | single | not recorded | decode_overlap | |
aggregate concurrency | 4.00 count | machine total | single | not recorded | aggregate_concurrency | |
requests total | 4.00 count | machine total | single | not recorded | requests_total | |
aggregate decode tps | 0.24 tok/s | machine total | single | not recorded | aggregate_tps | |
aggregate e2e tps | 0.18 tok/s | machine total | single | not recorded | aggregate_end_to_end_tps | |
peak host unavailable gib | 82.89 GiB | machine total | max | not recorded | peak_host_unavailable_gib |
Data and citation
Skilak Keep evidence, run 23cac8cdc0f4402ba860d186, measured 2026-10-06T13:19:49.255552+00:00. https://skilak.ai/keep/resources/evidence/r/23cac8cdc0f4402ba860d186
