Fetching from the wire…
01
02
03
04
05
06
07
08
Research2026-08-20 · source-backed
Across five TTS methods and five benchmarks spanning medicine, law, finance, chat and creative writing: candidate generation kept improving with compute in every domain, but reward models correlated with actual quality at roughly ρ=0.12. Only candidate fusion consistently beat single-sample, recovering about 40% of available quality gain. (arXiv 2608.18931) So best-of-N with a reward-model picker is a bad deal on subjective tasks. Synthesize across candidates instead of ranking them.
Each link below shares sources, entities, or timing with this story.
Shared entity: Test / Same source domain / Shared topic / Earlier coverage / Tension
Both cover Test; reported by the same outlet (arxiv.org); overlapping topics (benchmark, gain).
Shared entity: Test / Same source domain / Earlier coverage / Tension / Downstream implication
Both cover Test; reported by the same outlet (arxiv.org); earlier Test coverage from 2026-07-30.
Shared entity: Test / Same source domain / Shared topic / Earlier coverage
Both cover Test; reported by the same outlet (arxiv.org); overlapping topics (beat, benchmark).
Both cover Test; reported by the same outlet (arxiv.org); overlapping topics (benchmark, compute).
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (beat, benchmark, compute, consistently); pushes against this story (against).
Shared entity: Test / Same source domain / Earlier coverage / Tension
Both cover Test; reported by the same outlet (arxiv.org); earlier Test coverage from 2026-08-11.
Shared entity: Only / Same source domain / Earlier coverage / Tension
Both cover Only; reported by the same outlet (arxiv.org); earlier Only coverage from 2026-07-28.
Both cover Only; reported by the same outlet (arxiv.org); earlier Only coverage from 2026-07-21.