Parallel cut research time and cost in half with GPT‑6 Astra

Parallel, a developer of AI agent infrastructure for knowledge work, reports that its agents using GPT‑6 Astra completed labor-market research tasks in half the time and at half the cost compared to prior models. The improvement addresses a long-standing challenge where high-quality research required larger models with extended reasoning, increasing time and resource consumption. GPT‑6 Astra’s efficiency gains stem from more focused searches and fewer steps to reach actionable results, streamlining the research process for Parallel’s workflows.

In a specific test, Parallel tasked its agent with researching six labor-market statistics across four states over six months. The agent autonomously searched multiple websites, collected data, and compiled findings into a single report. GPT‑6 Astra completed the task in half the time of prior models while maintaining the same research quality, demonstrating tangible efficiency gains in real-world use.

Parallel also noted that GPT‑6 Astra reduced code costs by approximately 50%, further lowering operational expenses for research tasks. The model’s ability to delegate subtasks to sub-agents enabled parallel processing, allowing multiple research threads to run simultaneously rather than sequentially. This shift improved throughput and reduced idle time between steps in complex research workflows.

The efficiency gains position Parallel to scale demanding research tasks more effectively, with less waiting time and lower costs per project. The company highlights GPT‑6 Astra as a practical solution for turning complex research questions into actionable insights faster and at a reduced expense, supporting broader adoption of AI-driven knowledge work.