Fetching from the wire…
01
02
03
04
05
06
07
08
Research2026-07-11 · source-backed
This study investigates dropping speculative decoding's lossless guarantee without any training, quantifying speed-ups against controlled capability drift. Standard spec decoding exactly preserves the sampling distribution. Relaxing it buys latency at a small distributional cost that's sometimes worth it. Paired with "Resample or Reroute?" on budget-aware model routing, the theme is that inference-time knobs, not just model choice, decide your cost/quality point.
Each link below shares sources, entities, or timing with this story.
ServiceNow deprecates Standard / Shared entity: Standard / Earlier coverage
Linked by a graph relationship (ServiceNow deprecates Standard); both cover Standard; earlier Standard coverage from 2026-07-10.
Shared entity: Standard / Same source domain / Earlier coverage / Tension
Both cover Standard; reported by the same outlet (arxiv.org); earlier Standard coverage from 2026-06-20.
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (against, cost, model); pushes against this story (against).