Cekura Bench
Cekura Bench is Compare 19 speech-to-text models and APIs on word error rate, streaming latency, and finalization speed across Pipecat and Ocular datasets.. It is ranked #274 on the AI Radar, in Voice AI, first seen 44 days ago and shared in 3 posts (4.3k views).
Speech-to-Text Benchmarks: STT 2026 Leaderboard Bench Converse-STT benchmark Results Source Accuracy vs. speed Lower and further left is better Best quadrant Best picks Final-text delay P50 (ms) Overall WER (%) Who wins what The leader on each measure Most accurate Reson8 2.38% Overall WER Real calls Reson8 2.93% Ocular WER Fastest final text Soniox STT-RT v5 38ms TTFS P50 Full leaderboard A caller stops talking. Soniox STT-RT v5 has the transcript ready in 38ms. Reson8 takes 235ms, but gets the fewest words wrong. Last updated Sep 16, 2026 · Methodology by Dileep Chagam · With Ocular Leaderboard Overall Overall WER Pipecat Pipecat WER…
What people said about Cekura Bench on X
A big benchmarks release from @cekuraAi today. Good benchmarks are very difficulat develop for a bunch of reasons: - Automated testing of complex, multi-turn, conversational systems is a fundamentally hard problem. Do you use an LLM to simulate a user? That's the most common approach, these days. But how do you…
— @kwindla, 14 h ago · 37 likes · see the post
Love the examples of voice use !
— @andimarafioti, 4 h ago · 3 likes · see the post
Awesome job @tarush_agarwal_ and the @cekuraAi team!
— @ConstanceGriso, 20 h ago · 0 likes · see the post
Alternatives to Cekura Bench
- Soniox | Speech-to-Text AI — Generate the same speech with multiple text-to-speech providers and hear the difference. Compare voices,…
- Dessa — Hear it for yourself
- ElevenLabs — Apply to join
- nvidia-nemotron-3.5-transcribe-30b-a4b — Automatic Speech Recognition model with 34B parameters, built with NeMo. Download, explore files and run it on Hugging…
- phonon-tts — Gradium’s post
- hermes-voice — the code, the setup skill and the "hey hermes" wake word
Cekura Bench in numbers
- Rank on the AI Radar: #274 of 10395
- Shared in 3 posts by 3 accounts: @kwindla, @andimarafioti, @ConstanceGriso
- 4.3k views on those posts
- First seen 44 days ago, last shared 4 h ago
- Pricing seen by AGK Intelligence: not stated
- Market: Voice AI
FAQ
What is Cekura Bench?
Compare 19 speech-to-text models and APIs on word error rate, streaming latency, and finalization speed across Pipecat and Ocular datasets. It was first shared on X 44 days ago and is ranked #274 on the AI Radar.
Is Cekura Bench free?
Pricing is not stated on the page we read.
Who shared Cekura Bench?
3 accounts on X, including @kwindla, @andimarafioti, @ConstanceGriso, in 3 posts totalling 4.3k views.
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 65.7k posts from 5k X accounts over the last 14 days, 10.4k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 14:10 UTC. Full method.