Report · resources

Rolling chat

Qwen3.6-35B-A3B · NVFP4 variant · NVIDIA NGC vLLM 0.29

Measured results / Rolling chat / Run b2acdc4b13a24c219a0f2641

Measured 2026-10-07 15:55 UTC · Record date 2026-10-07

Failed. 0/50 turns

Answer checks

0/50 turns

Rolling-chat turn checks
All rolling-chat turns
Person · turnFilled inputFirst wordRecent · pinned · oldResultRun
1 · 1
null
not recorded s
false · false · invalid_response
Failed
1 · 2
null
not recorded s
false · false · invalid_response
Failed
1 · 3
null
not recorded s
false · false · invalid_response
Failed
1 · 4
null
not recorded s
false · false · invalid_response
Failed
1 · 5
null
not recorded s
false · false · invalid_response
Failed
1 · 6
null
not recorded s
false · false · invalid_response
Failed
1 · 7
null
not recorded s
false · false · invalid_response
Failed
1 · 8
null
not recorded s
false · false · invalid_response
Failed
1 · 9
null
not recorded s
false · false · invalid_response
Failed
1 · 10
null
not recorded s
false · false · invalid_response
Failed
2 · 1
null
not recorded s
false · false · invalid_response
Failed
2 · 2
null
not recorded s
false · false · invalid_response
Failed
2 · 3
null
not recorded s
false · false · invalid_response
Failed
2 · 4
null
not recorded s
false · false · invalid_response
Failed
2 · 5
null
not recorded s
false · false · invalid_response
Failed
2 · 6
null
not recorded s
false · false · invalid_response
Failed
2 · 7
null
not recorded s
false · false · invalid_response
Failed
2 · 8
null
not recorded s
false · false · invalid_response
Failed
2 · 9
null
not recorded s
false · false · invalid_response
Failed
2 · 10
null
not recorded s
false · false · invalid_response
Failed

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. No summaries. 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

  • 0/50 requests passed.

What ran

Exact identity

Original publisher model
Qwen3.6-35B-A3B
Publisher revision
995ad96eacd98c81ed38be0c5b274b04031597b0
Variant · format · precision
NVFP4 variant · safetensors · NVFP4
Weights SHA-256
not recorded
Engine · version · feature
NVIDIA NGC vLLM · 0.29.0+5013de39.dev · plain
Hardware
NVIDIA GB10, 128 GB unified memory
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
5
Configured window (tokens)
262144
Configured window (tokens)
262144
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
requests ok
0.00 count
machine total
single
not recorded
requests_ok
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
rolled off correct
0.00 count
machine total
single
not recorded
rolled_off_correct
rolled off total
0.00 count
machine total
single
not recorded
rolled_off_total
requests total
50.00 count
machine total
single
not recorded
requests_total

Data and citation

Skilak Keep evidence, run b2acdc4b13a24c219a0f2641, measured 2026-10-07T15:55:39.656970+00:00.
https://skilak.ai/keep/resources/evidence/r/b2acdc4b13a24c219a0f2641

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.