Parallel's gpt-6 astra cuts research time and costs in half

Parallel, a company that provides developer tools for AI agents performing knowledge work, has announced that its new model, GPT-6 Astra, significantly reduces the time and cost of research tasks. In a test case, the model completed a complex labor-market research project in half the time and at 50% lower cost compared to previous models. The task involved analyzing six labor-market statistics across four states over six months, requiring the agent to gather data from multiple websites and compile it into a report. GPT-6 Astra also demonstrated more focused searches and fewer steps to reach a useful result. Parallel noted that the increased efficiency allows for more practical division of research among multiple agents, enabling parallel processing and reducing overall time. This advancement provides a clearer path for tackling complex research questions with lower costs and greater scalability. The model's efficiency is attributed to its ability to delegate specific tasks to sub-agents, allowing simultaneous work and reducing the need for sequential searches. Parallel highlighted that this improvement makes it more feasible to handle demanding research tasks at scale, with less waiting time and reduced costs. The company emphasized that GPT-6 Astra maintains the same quality of research while achieving these efficiency gains.