Industry·
Asana Achieves 76x Cost Cut and 5x Speed Boost in Browser Agent Using GPT-6 Astra
In recent tests, Asana’s browser agent powered by GPT-6 Astra in Codex reduced operating costs by 76 times while increasing speed fivefold, demonstrating major gains for AI‑driven workflow automation.
Asana integrated the GPT-6 Astra model into its Codex‑based browser agent to evaluate cost and latency improvements.
The experiments showed that the agent’s inference expenses dropped by a factor of 76 compared with the previous setup, while response times became five times faster.
Why it matters for GPU / AI infrastructure
Such dramatic cost reductions highlight how newer model variants can achieve far higher throughput on existing GPU clusters, lowering the total cost of ownership for AI‑powered services.
For infrastructure planners, the results suggest that investing in optimized inference stacks and newer model releases can deliver substantial savings without sacrificing performance.
- aigpu
- ai gpu
- ai gpu cloud
- aigpu dubai
- gpt-6
- browser agent
- cost efficiency
By AiGpu Editorial · Editorial rewrite based on public reporting (OpenAI Blog)
← All articles