According to OpenAI News, conversational AI platform Retell is now utilizing OpenAI models to support a volume of 100 million monthly calls. Benchmark testing from the company indicates that GPT-6 Astra achieved top performance for its voice agent framework. Additionally, the integration of Codex has contributed to operational efficiency, helping Retell achieve an annual revenue run rate of 1.6 million dollars per employee.
The Role of Voice Agents in Enterprise
Voice agent infrastructure has become an increasingly competitive sector within artificial intelligence. Companies rely on large language models to reduce latency, manage conversational context, and execute complex workflows over telephone and VoIP channels. The integration of advanced coding and reasoning models allows platforms like Retell to deploy and scale automated systems capable of handling high call volumes without a proportional increase in human overhead.
What Remains Unknown
While the overall call volume and revenue metrics provide a snapshot of the company’s scale, the announcement did not specify the exact breakdown of which OpenAI models are handling specific workloads outside of GPT-6 Astra and Codex. Furthermore, the report did not disclose the timeline for these milestones or detail the underlying infrastructure costs required to maintain 100 million monthly calls.
As voice AI deployment accelerates, industry observers will monitor how infrastructure providers manage reliability and cost at this scale.
Lead image: RDNE Stock project / Pexels
