OpenAI says Retell runs 100M monthly calls with GPT-6 Astra atop its voice benchmark
OpenAI released a customer case study for Retell AI that positions the voice-agent platform as operating at very large scale on OpenAI models. According to the post, Retell uses OpenAI technology to help power roughly one hundred million monthly calls, and OpenAI says its GPT-6 Astra model ranks first on Retell's internal voice-agent benchmark. The same write-up highlights developer productivity, crediting Codex with contributing to a reported one point six million dollar annual revenue run rate per employee. For teams building production voice agents, the figures offer a rare vendor-published snapshot of call volume and model positioning, though the metrics come from OpenAI's own marketing case study rather than independent verification. Separately, industry coverage in the same signal window touches enterprise pricing shifts and UK biosecurity funding involving OpenAI, but those threads are not detailed in the Retell post itself.
OpenAI says Retell runs 100M monthly calls with GPT-6 Astra atop its voice benchmark
OpenAI published a Retell case study stating Retell uses OpenAI models to help power 100M monthly calls. The post also says GPT-6 Astra tops Retell’s voice agent benchmark and credits Codex with helping Retell reach a $1.6M annual revenue run rate per employee.
Key takeaway
OpenAI's Retell case study links GPT-6 Astra to a top voice benchmark score and cites 100M monthly calls, but the scale and revenue metrics are self-reported in vendor marketing.
What happened
OpenAI News published a Retell case study stating that Retell uses OpenAI models to help power 100M monthly calls, framing the deployment as a large-scale voice-agent workload on OpenAI infrastructure.
The same OpenAI post says GPT-6 Astra tops Retell's voice agent benchmark and credits Codex with helping Retell reach a $1.6M annual revenue run rate per employee, per the attributed case-study language in the packet.
Evidence
Retell uses OpenAI models to help power 100M monthly calls.
OpenAI News · attributed
Retell uses OpenAI models to help power 100M monthly calls
GPT-6 Astra tops Retell's voice agent benchmark.
OpenAI News · attributed
GPT-6 Astra tops Retell's voice agent benchmark, while Codex helps it reach a $1.6M annual revenue run rate per employee.
A source told The Information that Replit moved most AI spend to OpenAI after Anthropic discontinued its discount, with 30%+ off in a deal worth $100M+/year.
Techmeme · attributed
Source: Replit recently moved most AI spend from Anthropic to OpenAI, which gave 30%+ off in a deal worth $100M+/year, after Anthropic discontinued its discount
The UK plans to invest £100M to create an AI Biosecurity Lab with support from OpenAI, Anthropic, and Google DeepMind.
Techmeme · attributed
The UK plans to invest £100M to create the AI Biosecurity Lab to boost defenses against biological threats, supported by OpenAI, Anthropic, and Google DeepMind
Eric Schmidt and co-authors argue in Foreign Affairs that the AI race may hinge on who can bankroll it, with US corporations planning more than $600 billion in spend in 2026.
The Decoder · attributed
The race won't be decided by breakthroughs but by who can bankroll it. US corporations alone plan to spend more than $600 billion in 2026.
Why it matters
Production voice stacks and enterprise buyers may treat vendor case studies as competitive signals even when independent benchmarks and third-party spend data remain thin or excerpt-only.
Limits and uncertainties
The Retell scale, benchmark, and per-employee revenue figures come from an OpenAI-published case study without independent verification in the packet.
Techmeme excerpts on Replit and the UK AIBL do not establish full contract terms, lab timelines, or operational deliverables.
The Decoder coverage of the Foreign Affairs piece is excerpt-only and does not establish the full argument or evidence behind the $600 billion figure.
Practical implications
Voice-agent operators should separate vendor benchmark claims from their own eval harness before committing model routes at scale.
Teams negotiating model spend should note reported discount-driven vendor switches while treating single-source deal values as unconfirmed.
AI-bio builders should wait for AIBL procurement and governance details before assuming a new UK funding channel is open.
What to watch
Whether OpenAI or Retell publish independent verification or technical detail behind the 100M monthly calls and voice benchmark methodology.
Follow-on reporting on Replit's model mix and contract terms after the reported Anthropic-to-OpenAI shift.
UK government releases on AIBL mandate, timeline, and how OpenAI, Anthropic, and Google DeepMind participate beyond the announcement.
