- Writing speed, one personsingle_user_decode_tps · tok/s
- 2.57 tok/sper person · single · sample count not recorded
- First word afterttft_median_s · s
- 1,282.11 sper person ·
- Slowest first wordttft_max_s · s
- 1,282.11 sper person ·
- Peak memory, whole hostpeak_host_unavailable_gib · GiB
- 12.87 GiBmachine total · max · sample count not recorded
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 | 112 | 1,282.11 s | 2.57 tok/sone person · includes waits | 1,325.38 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 | 112 | 1,282.11 s | 2.57 tok/sone person · includes waits | 1,325.38 s | 3/3 Hidden facts found · completed · checks passed |
Answer checks
3/3 Hidden facts found1/1 request checks
Method
One cold 250000-content-token document; original tokenizer cross-checked natively; strict JSON three exact codes and synthesis presence; no retries.
Limits
Correct retrieval and completion; first token exceeds the 300-second reporting bar. Configured window is 262144, not a filled 262144-token input; synthesis semantics and broader quality unassessed.
What ran
Exact identity
- Original publisher model
- Qwen3.8-27B
- Publisher revision
not recorded- Variant · format · precision
- Mirai S trellis with Q8_0 MTP GGUF variant · GGUF · Mirai S trellis with Q8_0 MTP
- Weights SHA-256
- not recorded
- Engine · version · feature
- Mirai S CUDA ·
0.2.0-dev-vector-disabled· with multi-token prediction - Hardware
- NVIDIA RTX 3060, nominal 12 GB, one GPU
- Harness / workload
not recorded- Workload suite · version
not recorded- Harness SHA-256
not recorded- Fixture SHA-256
not recorded
Workload shape
- cache state
- cold
- People in this run
- 1
- Filled content target (tokens)
- 250000
- fixture variant
- original-tokenizer-needles-v1
- KV cache precision
- Q8_0
- Output token ceiling
- 512
- scenario
- distinct
- Feature
- with multi-token prediction
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 | 1,282.11 s | per person | median | 1 | ttft_median_s | |
hidden facts total | 3.00 count | machine total | single | not recorded | needle_total | |
ttft max s | 1,282.11 s | per person | max | 1 | ttft_max_s | |
requests ok | 1.00 count | machine total | single | not recorded | requests_ok | |
hidden facts correct | 3.00 count | machine total | single | not recorded | needle_correct | |
latency bar seconds | 300.00 s | machine total | single | not recorded | latency_bar_seconds | |
decode overlap | 1.00 count | machine total | single | not recorded | decode_overlap | |
requests total | 1.00 count | machine total | single | not recorded | requests_total | |
single user decode tps | 2.57 tok/s | per person | single | not recorded | single_user_decode_tps | |
peak host unavailable gib | 12.87 GiB | machine total | max | not recorded | peak_host_unavailable_gib |
Data and citation
Skilak Keep evidence, run caf8d447ceff44c9bd32727f, measured 2026-10-08T04:24:25.168145+00:00. https://skilak.ai/keep/resources/evidence/r/caf8d447ceff44c9bd32727f
