Retell Powers 100M Monthly Calls With OpenAI Models

Voice AI platform Retell AI uses OpenAI models to power more than 100 million monthly calls for over 4,000 businesses, according to an update detailing the platform's recent growth and technical benchmarks.

Monthly call volume on the platform grew from 30 million in early 2026 to more than 100 million by September. OpenAI models currently generate responses for 80% of calls handled by Retell, while customers achieve customer resolution rates above 80%. The platform has surpassed 1 trillion tokens consumed across OpenAI models.

To evaluate model performance for live conversations, Retell runs its vCX benchmark. In testing conducted on October 13, 2026, GPT-6 Astra ranked first for instruction following at 89.7%. Pre-launch evaluations showed GPT-5.5 ranking highest on non-hallucination rate at 88.0% and tool-call correctness at 88.3% with reasoning enabled, outperforming Gemini-3.1-Pro and Claude-4.6.

For latency, GPT-6 Astra and GPT-6 Sol each recorded a median first-sentence latency of 883 milliseconds in the October 13 benchmark, while GPT-5.6 Luna ranked highest at 497 milliseconds.

Enterprise deployments include electronics company Anker, which reported an 80.4% case resolution rate and a Net Promoter Score of 63 using Retell agents for Tier-1 post-sales support in the US and UK. Pine Park Health used agents for appointment reminders and scheduling, recovering medical-assistant time equivalent to 2.1 full-time roles and achieving a 55.7% booking rate.

Retell engineers use Codex throughout development for tasks such as researching code and debugging issues. The platform also employs GPT-6 Astra's computer use capabilities to test front-end web interfaces. Retell reports an annual revenue run rate per employee of $1.6M.