AssemblyAI's Voice Standard Bench ranks 50+ LLMs on real speech tasks
AssemblyAI · x · 2026-10-14
AssemblyAI launched Voice Standard Bench, ranking 50+ models on its LLM Gateway by quality, speed and cost on real Speech-to-Text output. For up to 15 minutes of audio, a model within two quality points of the leader costs 50–85× less.
Highlights (0–100 score):
- Highest performance: GPT-6 Astra, 94
- Lowest cost: Nemotron 3 Nano 30B A3B, $0.003/hour of audio
- Fastest: Qwen3.5 4B Fast, 0.60s median response
Tasks covered include information extraction, summarization, text correction and dictation cleanup.
Related event: AssemblyAI Launches Voice Standard Bench for 50+ Models(5 posts)→
More from Models
- Gemini 4 Argon lands day-0 on Databricks, tops internal coding and agentic benchmarks — matei_zaharia · 2026-10-14
- Argon tops hallucination bench with >50% false claim rate, prompting Gemini users to back off — brandon_galang · 2026-10-14
- After a sudden ban, this Claude user stopped treating AI accounts as assets — gefei55 · 2026-10-14
- Gemini disables Google Search once you attach a document, user discovers — dejanseo · 2026-10-14
- Reddit users complain Gemini has gotten noticeably dumber — Prestigious_Use5944 · 2026-10-14
- Claude Fable 5.1 vs Muse Wonder 1.0 benchmark draws Alexandr Wang reaction — alexandr_wang · 2026-10-14