Parallel Cuts Research Time and Cost in Half with GPT-6 Astra
OpenAI, Tuesday, September 22nd, 2026
Research infrastructure firm Parallel says GPT-6 Astra cut its agents' research time and cost roughly in half versus prior models.
Parallel, which builds developer infrastructure for AI agents doing web-based knowledge work, tested GPT-6 Astra on a research task requiring an agent to gather six labor-market statistics across four states over six months and compile a report.
Astra completed the work in half the time of prior models with roughly a 50% cost reduction while maintaining research quality, taking more focused searches and fewer steps to reach a useful result.
The efficiency gains also make it more practical for Parallel to divide research tasks among multiple sub-agents working simultaneously.