SkilakKeep
Request a pilot

Report · resources

Chat and code

Qwen3.8-27B · Mirai S trellis with Q8_0 MTP GGUF variant · Mirai S CUDA dev · with multi-token prediction

Measured results / Chat and code / Run 68a30d35b8064ec9b775f2d0

Measured 2026-10-09 14:21 UTC · Record date 2026-10-09

Not graded. 0/0 Hidden facts found1/1 request checks

Prose writing speedchat_decode_tps · tok/s
33.08 tok/sper person · mean · sample count not recorded
Peak memory, whole hostpeak_host_unavailable_gib · GiB
12.58 GiBmachine total · max · sample count not recorded

Answer checks

0/0 Hidden facts found1/1 request checks

Method

Frozen original-tokenizer fixture; exact native count verified; owned managed container stopped and recreated before dispatch. Content tokens 32000; actual prompt tokens 32012; decode uses output tokens minus one divided by first-content-to-end time; no retries.

Limits

Depth prose measures speed, not complete answers. hidden facts JSON retrieval is finite evidence; synthesis semantics are unassessed. Whole-host memory includes unrelated services. Managed lab evidence, not customer acceptance.

What ran

Exact identity

Original publisher model
Qwen3.8-27B
Publisher revision
1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0
Variant · format · precision
Mirai S trellis with Q8_0 MTP GGUF variant · GGUF · Mirai S trellis with Q8_0 MTP
Weights SHA-256
af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334
Engine · version · feature
Mirai S CUDA · 0.2.0-dev-vector-disabled · with multi-token prediction
Hardware
NVIDIA GeForce RTX 3060, 12 GB VRAM
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
Configured window (tokens)
262144
Configured window (tokens)
262144
KV cache precision
q8_0
Output token ceiling
512
Feature
with multi-token prediction
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
hidden facts total
0.00 count
machine total
single
not recorded
needle_total
chat decode tps
33.08 tok/s
per person
mean
not recorded
chat_decode_tps
requests ok
1.00 count
machine total
single
not recorded
requests_ok
hidden facts correct
0.00 count
machine total
single
not recorded
needle_correct
latency bar seconds
300.00 s
machine total
single
not recorded
latency_bar_seconds
requests total
1.00 count
machine total
single
not recorded
requests_total
chat ttft mean s
66.87 s
per person
mean
not recorded
chat_ttft_mean_s
peak host unavailable gib
12.58 GiB
machine total
max
not recorded
peak_host_unavailable_gib

Data and citation

Download this run · OLIR JSONAll numbers · CSVOLIR schema
Skilak Keep evidence, run 68a30d35b8064ec9b775f2d0, measured 2026-10-09T14:21:18.048632+00:00.
https://skilak.ai/keep/resources/evidence/r/68a30d35b8064ec9b775f2d0

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.

Request a pilot