In a five-draft/one-verify cycle on Qwen3-8B, the refresh layer added 1.218 ms (1.03% of cycle time) at 16k context and 2.113 ms (1.06%) at 32k. On Qwen3-1.7B and DS-Llama-8B, overhead was at or below the measurement-noise floor.

In a five-draft/one-verify cycle on Qwen3-8B, the refresh layer added 1.218 ms (1.03% of cycle time) at 16k context and 2.113 ms (1.06%) at 32k. On Qwen3-1.7B and DS-Llama-8B, overhead was at or below the measurement-noise floor. · CiteArk

Measurement results

0/4 assessed

0 supported0 disputed or conflicting4 not assessed

The complete result group stays with its claim. Assessed rows come first; their judgments do not replace the whole-claim conclusion. Expand a row for conditions, sources, and evidence history.

Each row is a measurement with its reported value, latest available observation, and cumulative evidence state.
Metric and conditions	Reported	Latest observation	Evidence judgment
model: Qwen3-8B · contextLengthTokens: 16000	1.218 ms/cycle	—	Not assessed
model: Qwen3-8B · contextLengthTokens: 16000	1.03%	—	Not assessed
model: Qwen3-8B · contextLengthTokens: 32000	2.113 ms/cycle	—	Not assessed
model: Qwen3-8B · contextLengthTokens: 32000	1.06%	—	Not assessed

In a five-draft/one-verify cycle on Qwen3-8B, the refresh layer added 1.218 ms (1.03% of cycle time) at 16k context and 2.113 ms (1.06%) at 32k. On Qwen3-1.7B and DS-Llama-8B, overhead was at or below the measurement-noise floor.

Reported and observed measurements

Assessments (0)

Runs (0)

Measurement results