GPT-6 Astra halves research time and costs for AI agents, OpenAI reports

Parallel, a developer infrastructure provider for AI agents, uses its tools to support various knowledge work, including web grounding and specialized research for financial and legal sectors. The company noted that while complex research tasks previously required larger models with extended reasoning—consuming significant time and resources—the implementation of GPT-6 Astra has brought major improvements in efficiency.

In a specific test, Parallel's agent was tasked with researching six different labor-market statistics across four states over a six-month period. The agent performed tasks such as searching multiple websites, collecting information, and compiling findings into a single report. GPT-6 Astra completed this work in half the time required by previous models and achieved a roughly 50% reduction in code cost while maintaining the same research quality.