# Retell uses OpenAI models to help power 100M monthly calls

Source: [OpenAI](https://openai.com/index/retell-ai)  
Feed7 permalink: https://feed7.dev/p/retell-ai-0dja9xg  
Published: 2026-10-27T15:00:00.000Z  
Trust: Official Source (official_source)

## Why Included

Retell says GPT-6 Astra leads its voice-agent benchmark, and that Codex contributes to a $1.6M annual revenue run rate per employee. The material omits benchmark method and attribution details.

## Source Summary

Retell reports that **GPT-6 Astra** tops its voice agent benchmark. The company also reports **$1.6M annual revenue run rate per employee** and credits Codex with helping it reach that figure.

## Practical Implication

Builders can treat the benchmark result as a prompt to evaluate voice models against their own tasks; the revenue figure may invite questions about how coding agents fit into company operations.

## Agent-Ready Context

Retell reports that **GPT-6 Astra** tops its voice agent benchmark. The company also reports **$1.6M annual revenue run rate per employee** and credits Codex with helping it reach that figure.

Builders can treat the benchmark result as a prompt to evaluate voice models against their own tasks; the revenue figure may invite questions about how coding agents fit into company operations.

The material does not describe the benchmark setup or isolate Codex's contribution to revenue, so neither claim supports a direct performance or productivity comparison.

## Connected Context

Feed7 judgment across 944 accumulated Signals:

Retell reports a leading GPT-6 Astra result on its voice-agent benchmark and high revenue per employee, crediting Codex for helping reach that figure. Without benchmark details or evidence isolating Codex’s contribution, these claims motivate evaluation but do not establish comparative model performance or a causal productivity gain.

- [Benchmarks: The Good, the Bad, and the Ugly — Ali Khial, G2i](https://feed7.dev/p/benchmarks-the-good-the-bad-and-the-ugly-ali-khial-g2i-0p1hw71) — The benchmark critique underscores why Retell’s unspecified setup and verification matter before treating its model ranking as transferable evidence.
- [Separating signal from noise in coding evaluations](https://feed7.dev/p/separating-signal-from-noise-coding-evaluations-17ha3r5) — Reported evaluation reliability problems in coding benchmarks reinforce the need to inspect Retell’s benchmark methodology before relying on its ranking.
- [How Forward Deployed Engineering is done at Cognition — Jia Wu](https://feed7.dev/p/how-forward-deployed-engineering-is-done-at-cognition-jia-wu-06h8ybj) — Cognition’s delivery-focused measures contrast with Retell’s revenue-per-employee figure, which does not isolate Codex’s contribution to operational outcomes.

## Context Map

- Layer: benchmark
- Domains: audio
- Topics: agent-evals, model-selection, coding-agents

## Uncertainty

- The material does not describe the benchmark setup or isolate Codex's contribution to revenue, so neither claim supports a direct performance or productivity comparison.

## Agent Instruction

Use this item as source-backed context. Do not invent claims beyond the linked source. If this item conflicts with another source, call out the conflict.