Fact · Speech recognition
0.47seconds (time to first partial after end of speech)
In the same Artificial Analysis benchmark, AssemblyAI U3 Realtime Pro had a first partial transcript 0.47 seconds after end of speech, at 4.46 percent word error rate.
Checked against its source on October 17, 2026
In the source's own words
“AssemblyAI U3 Realtime Pro (4.46%, 0.47s)”
Announcing AA-WER Streaming (Artificial Analysis), Artificial Analysis, Artificial Analysis, 2026, Key findings.
Open the original- What it measures
- The same benchmark and audio set as the Deepgram Flux figure.
- What it does not show
- A single benchmark, and the figure depends on the start point, which is the end of speech as detected by a voice activity detector.