Perplexity has started using the GPT‑6 Astra model to carry out system-level tasks such as drafting communications, making software changes, and monitoring production systems. The company reports that Astra checks in far less frequently than earlier models while continuing to handle large volumes of information.
Johnny Ho, Cofounder and Chief Strategy Officer of Perplexity, noted that improvements in the model’s coding ability directly benefit Perplexity’s search engine: as the model produces better code, the engine becomes more capable of searching the web and internal data sources and summarizing findings concisely.
According to Ho, the larger challenge is applying those informational capabilities to real-world systems — an area where GPT‑6 Astra is particularly helpful. Perplexity gives the model responsibilities that go beyond answering questions, using it to interact with and influence operational systems.
One practical use case Ho highlighted is code testing. With limited time for manual testing, he asks GPT‑6 Astra to build a small testing program around an application. The model can generate realistic responses that mimic other services — for example, a language model API or a connector — and by standing in for those services it can exercise the application’s behavior and validate end‑to‑end workflows.
In short, Perplexity is deploying GPT‑6 Astra not only for information processing but also for automated, system-level tasks such as testing and service simulation, where the model can emulate external components and test full workflows.
