A report from OpenAI News says that when Asana’s browser agent leveraged GPT-6 Astra within Codex, the tests showed the agent becoming 76x cheaper and 5x faster. The testing context notes the objective of offering customers more capable models.
WHAT CHANGED IN THE TESTS
The described testing setup uses GPT-6 Astra in Codex to optimize the browser agent’s performance, resulting in substantially lower costs and higher speed in the browser environment. TechStaged has also covered Meta’s Muse Spark Shows the Race Is Moving From Chatbots to AI Products.
WHY THE DEVELOPMENT MATTERS FOR CUSTOMERS AND THE MARKET
If these test results translate to production environments, customers could see more capable browser-based AI models at lower operating costs and with faster response times, according to the report describing the tests.
CONTEXT AND WHAT TO WATCH FOR NEXT
The claims come from a single open report about browser tests. The article notes the aim of delivering more capable models to customers, but it does not provide additional details about deployment timelines, broader product impact, or independent verifications.
RELATED COVERAGE
- Meta’s Muse Spark Shows the Race Is Moving From Chatbots to AI Products
- NVIDIA GPUs Accelerate OpenAI’s GPT-6 Astra Ultrafast
- NVIDIA Demonstrates Frontier AI Agents Turning Simulation Ideas Into Working Omniverse Apps
- Sophos cuts threat investigation time by 96% with OpenAI Daybreak
- AI & Automation articles
SOURCES
- OpenAI News: Asana cuts model costs 76x in browser tests with GPT-6.1 Sol Published · Primary source
