Retell says its platform handled more than 100 million monthly calls by September, with OpenAI models generating responses for 80% of calls.
Retell AI says its voice-agent platform surpassed 100 million calls in a month by September 2026, up from 30 million monthly calls early in the year. In an October 27 account of the partnership, OpenAI said Retell serves more than 4,000 businesses and has used over 1 trillion tokens across OpenAI models.
OpenAI models generate responses for 80% of calls on the platform, according to the company, while Retell reports customer resolution rates above 80%. Retell’s own October 13 benchmark of models on challenging, multi-turn voice interactions ranked GPT-6 Astra first for instruction following, at 89.7%. The benchmark tests instruction adherence, grounding, tool use and conversational language. In separate evaluations conducted before GPT-6 Astra’s launch, Retell reported that GPT-5.5 led on non-hallucination rate and tool-call correctness with reasoning enabled.
The article describes deployments at two customers. Anker uses Retell agents for Tier-1 post-sales support in the United States and United Kingdom; the company reported an 80.4% case-resolution rate and a Net Promoter Score of 63. Pine Park Health uses agents for appointment reminders and scheduling. Retell says the work returned medical-assistant time equivalent to 2.1 full-time roles, and the scheduling agent booked appointments in 55.7% of calls where it reached a patient.
For call responsiveness, Retell reports that daytime traffic can exceed 50 times its overnight low. Its October benchmark measured median first-sentence latency of 883 milliseconds for GPT-6 Astra and GPT-6 Sol, while GPT-5.6 Luna recorded 497 milliseconds. Retell also uses regional API endpoints and backup telephony, voice and model services.
Retell engineers use Codex for code research, implementation and debugging, and GPT-6 Astra’s computer-use capability to test web-interface changes. OpenAI says Retell reports an annual revenue run rate of $1.6 million per employee. The company plans to expand quality assurance and simulation, add language support and develop automatic knowledge-base learning.