Workplace software provider Asana has evaluated GPT-6 Astra inside Codex to power its browser agent tooling. According to benchmark figures published by OpenAI, Asana made its browser agent 76x cheaper and 5x faster in tests during these evaluations. The testing focused on measuring cost efficiency and execution latency for agent-driven workflows.
The primary operational objective behind these adjustments is to offer customers more capable models without running into prohibitive operational constraints. By executing tasks through GPT-6 Astra in Codex, the browser agent achieved significant gains in responsiveness while reducing testing costs. These performance gains reflect Asana's reported efforts to supply users with enhanced artificial intelligence capabilities.