DATASETS2SCOVAL BENCHMARK ROTATIONACTIVE

Customer service speech-to-speech benchmark dataset

Name: Customer service voice AI benchmark dataset
Creator: Coval benchmark rotation

Customer service is a speech-to-speech test set in Coval's daily rotation — every measured speech-to-speech model runs the same fixed inputs from it.

How models rank on Customer service

Instruction Adherence on Customer servicePercent · higher is better · 30-day average on this dataset

median of all models · 42%

S2S models on the Customer service dataset over the last 30 days, ranked on Instruction Adherence.
#	Model	Host	Instruction adherence	Samples
1	GPT Realtime 2	OpenAI	42%	48

Under-sampled models are excluded; tied models share a place.

Same datasets, prompts and metric definitions for every model, measured by Coval’s open-source runner. Full methodology on the overview.

Evaluate your own voice agent

Use Coval to test your production configuration, prompts and calls—not only the public benchmark endpoints.