Report · resources

Rolling chat

Qwen3.8-27B · MIRAI-S-TRELLIS-Q8-MTP variant · Mirai S CUDA dev

Measured results / Rolling chat / Run 1d3b3864eeed452b9756596f

Measured 2026-10-09 17:00 UTC · Record date 2026-10-09

Failed. 4/4 turns

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
2026-10-09
1 people
not recorded
not recorded
not recorded
Failed
2026-10-09
1 people
not recorded
not recorded
not recorded
Failed

Answer checks

4/4 turns

Rolling-chat turn checks
All rolling-chat turns
Person · turnFilled inputFirst wordRecent · pinned · oldResultRun
1 · 1
20211
40.06 s
true · true · recalled
Passed
1 · 2
40340
49.26 s
true · true · recalled
Passed
1 · 3
60470
62.26 s
true · true · recalled
Passed
1 · 4
40513
92.40 s
true · true · recalled
Passed

Method

Staggered multi-turn fictional archive, exact engine-tokenized rolling gateway; independent recent, pinned and old probes each turn.

Limits

Bounded tuning pilot; incomplete window growth cannot pass the full workload grade. Lab retrieval and rolling-history evidence only. Bounded literal user-source excerpts, with verified source hashes; details outside them remain unavailable. Shared engine cache intervals include overlapping users and are not per-user hits. Gateway path, not authenticated document ingestion or customer acceptance.

What did not work

  • Bounded tuning pilot; incomplete window growth cannot pass the full workload grade.

What ran

Exact identity

Original publisher model
Qwen3.8-27B
Publisher revision
1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0
Variant · format · precision
MIRAI-S-TRELLIS-Q8-MTP variant · gguf · MIRAI-S-TRELLIS-Q8-MTP
Weights SHA-256
af3e2d508cb12ebf68688a0756da6bcb4cdfceca3883f5751666d5e4fb34a334
Engine · version · feature
Mirai S CUDA · 0.2.0-dev-vector-disabled · plain
Hardware
NVIDIA GeForce RTX 3060, 12 GB VRAM
Harness / workload
keep-rolling-context-v1 · 1
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)
262144
Configured window (tokens)
64000
Output token ceiling
256

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
roll ttft max seconds
92.40 s
per person
max
not recorded
roll_ttft_max_seconds
requests ok
4.00 count
machine total
single
not recorded
requests_ok
roll ttft p50 seconds
92.40 s
per person
median
not recorded
roll_ttft_p50_seconds
rolled off abstained
0.00 count
machine total
single
not recorded
rolled_off_abstained
rolled off invalid
0.00 count
machine total
single
not recorded
rolled_off_invalid
rolled off wrong
0.00 count
machine total
single
not recorded
rolled_off_wrong
normal ttft max seconds
62.26 s
per person
max
not recorded
normal_ttft_max_seconds
rolled off correct
1.00 count
machine total
single
not recorded
rolled_off_correct
normal ttft p50 seconds
49.26 s
per person
median
not recorded
normal_ttft_p50_seconds
rolled off total
1.00 count
machine total
single
not recorded
rolled_off_total
requests total
4.00 count
machine total
single
not recorded
requests_total

Data and citation

Skilak Keep evidence, run 1d3b3864eeed452b9756596f, measured 2026-10-09T17:00:14.157340+00:00.
https://skilak.ai/keep/resources/evidence/r/1d3b3864eeed452b9756596f

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.