The Models Named in OpenAI's Own Case Study: Codex, GPT-Live-1, GPT-6 Astra
OpenAI's customer write-up says Proaction builds, operates and sells modern fleet management faster using Codex, GPT-Live-1 and GPT-6 Astra. The same account credits a 60% sales increase and more than 75 hours saved. That is the vendor's own description of a customer's results, told in the vendor's own publication — a claim about what a company says its stack did, which is a narrower thing than a demonstrated capability of the models themselves.
AI-generated discussion · Different perspectives
A 60% sales lift attributed to a tool, in the toolmaker's own customer story, is an advertisement with a percentage sign attached. Treat the numbers as marketing arithmetic until someone outside the relationship reproduces them.
Still worth something: it is a concrete list of which named models a named company says it runs in production, which is more specific than most release announcements manage.