Report · resources

Long documents

Qwen3.8-27B · INT4 variant · NVIDIA NGC vLLM 0.29

Measured results / Long documents / Run 641d1e53d16a453a8834e31f

Measured 2026-10-06 06:22 UTC · Record date 2026-10-06

Passed. 15/15 Hidden facts found5/5 request checks

First word afterttft_median_s · s
8.84 sper person · median of 5
Slowest first wordttft_max_s · s
10.70 sper person · max of 5
Machine totalaggregate_decode_tps · tok/s
16.13 tok/smachine total · single · sample count not recorded
Machine total including first-word waitaggregate_e2e_tps · tok/s
14.21 tok/smachine total · single · sample count not recorded
Everyone answered inwall_seconds · s
36.32 smachine total · single · sample count not recorded
Peak memory, whole hostpeak_host_unavailable_gib · GiB
104.31 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
History of this setup, newest first
DateShapeEach person · medianMachine totalFirst word · medianResultRun
2026-10-06
5 people · distinct · warm
not recorded
16.13 tok/s
8.84 s
Passed
2026-10-06
5 people · distinct · cold
not recorded
12.53 tok/s
2,287.02 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.

Each person's first word and writing speed; table below00 s26 s511 stok/s after first wordfirst word afterPerson 2: 4.63 s · 3.28 tok/sPerson 3: 8.84 s · 3.76 tok/sPerson 1: 8.84 s · 3.76 tok/sPerson 4: 10.70 s · 4.00 tok/sPerson 5: 10.70 s · 4.02 tok/s
Per-person table · all measured request rows
Per-person requests for this run
PersonFilled inputCached tokensOutputFirst wordWriting after first wordTotal timeChecksRun
2
250,011
unknown
103
4.63 s
3.28 tok/sone person · includes waits
35.70 s
3/3 Hidden facts found · completed · checks passed
3
250,011
unknown
103
8.84 s
3.76 tok/sone person · includes waits
35.97 s
3/3 Hidden facts found · completed · checks passed
1
250,011
unknown
103
8.84 s
3.76 tok/sone person · includes waits
35.97 s
3/3 Hidden facts found · completed · checks passed
4
250,011
unknown
103
10.70 s
4.00 tok/sone person · includes waits
36.17 s
3/3 Hidden facts found · completed · checks passed
5
250,011
unknown
104
10.70 s
4.02 tok/sone person · includes waits
36.31 s
3/3 Hidden facts found · completed · checks passed
Per-person table · all measured request rows
Per-person requests for this run
PersonFilled inputCached tokensOutputFirst wordWriting after first wordTotal timeChecksRun
2
250,011
unknown
103
4.63 s
3.28 tok/sone person · includes waits
35.70 s
3/3 Hidden facts found · completed · checks passed
3
250,011
unknown
103
8.84 s
3.76 tok/sone person · includes waits
35.97 s
3/3 Hidden facts found · completed · checks passed
1
250,011
unknown
103
8.84 s
3.76 tok/sone person · includes waits
35.97 s
3/3 Hidden facts found · completed · checks passed
4
250,011
unknown
103
10.70 s
4.00 tok/sone person · includes waits
36.17 s
3/3 Hidden facts found · completed · checks passed
5
250,011
unknown
104
10.70 s
4.02 tok/sone person · includes waits
36.31 s
3/3 Hidden facts found · completed · checks passed

Answer checks

15/15 Hidden facts found5/5 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 ran

Exact identity

Original publisher model
Qwen3.8-27B
Publisher revision
1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0
Variant · format · precision
INT4 variant · safetensors · INT4
Weights SHA-256
1d8268aa85ace093a561e3e7b63b9d390dac1cd55a90cd55b5ec509c3c9da9fe
f078f51aed0c1dba613c1759487004ee06aba1a290606b6be125f59ab710959e
Engine · version · feature
NVIDIA NGC vLLM · 0.29.0+5013de39.dev · plain
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
warm
People in this run
5
Configured window (tokens)
262144
Filled content target (tokens)
250000
fixture variant
distinct-v1
KV cache precision
fp8
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
All aggregate numeric result fields in this run
MetricValueScopeStatisticSample countBasis / source fieldRun
wall seconds
36.32 s
machine total
single
not recorded
wall_seconds
ttft median s
8.84 s
per person
median
5
ttft_median_s
hidden facts total
15.00 count
machine total
single
not recorded
needle_total
ttft max s
10.70 s
per person
max
5
ttft_max_s
requests ok
5.00 count
machine total
single
not recorded
requests_ok
hidden facts correct
15.00 count
machine total
single
not recorded
needle_correct
decode overlap
5.00 count
machine total
single
not recorded
decode_overlap
aggregate concurrency
5.00 count
machine total
single
not recorded
aggregate_concurrency
requests total
5.00 count
machine total
single
not recorded
requests_total
aggregate decode tps
16.13 tok/s
machine total
single
not recorded
aggregate_tps
aggregate e2e tps
14.21 tok/s
machine total
single
not recorded
aggregate_end_to_end_tps
peak host unavailable gib
104.31 GiB
machine total
max
not recorded
peak_host_unavailable_gib

Data and citation

Skilak Keep evidence, run 641d1e53d16a453a8834e31f, measured 2026-10-06T06:22:47.697268+00:00.
https://skilak.ai/keep/resources/evidence/r/641d1e53d16a453a8834e31f

Ready to scope it?

Start with one workload.

Tell us who uses it, what it can read, and what must stay inside. We will map the build.