Third-party benchmark: AssemblyAI Universal 3.5 Pro tops 15 STT models at 1.93% WER and 489ms latency
AssemblyAI · x · 2026-09-24
The third-party Converse-STT benchmark (Cekura Bench) ran 15 speech-to-text models through the same audio pipeline, and AssemblyAI Universal 3.5 Pro came first on both accuracy and latency.
- On 1,000 read-aloud clips from Pipecat FLEURS, it posted the lowest word error rate at 1.93%, ahead of Google Chirp 3 (2.00%), Reson8 (2.10%) and GPT Realtime Whisper (2.16%).
- Its median time to first text (TTFT P50) was 489ms, the fastest of any model tested — Google Chirp 3 took 4.48s on the same metric.
- Overall it sits on the quality-speed Pareto frontier, making it a strong pick for latency-sensitive voice agent pipelines.
Full results are public on Cekura Bench, with source code available.
More from Models
- Do typos discomfort LLMs? Tokenization quirk sparks debate — menhguin · 2026-11-11
- MiniMax H3 hits 100 days open-source: 1,472 derivative models, 148M+ downloads — linoy_tsaban · 2026-11-11
- Meituan's LongCat-2.5 trends on Hugging Face as MoE multimodal model — meituan-longcat · 2026-11-11
- Fable 5.2 hands-on: great at well-defined hard problems, best since GPT 6 Astra — rachittshah · 2026-11-11
- The next AI vendor lock-in won't be the model API, it'll be the harness — Arindam_1729 · 2026-11-11
- Leaked: OpenAI's GPT-6.2 (codename betelgeuse) reportedly weighs in at ~15T tokens, drops Thursday — iruletheworldmo · 2026-11-11