Skip to content

Back to the catalog

openai.comarticle

Asana cuts model costs 76x in browser tests with GPT-6.1 Sol

Asana's StackAI browser agent became 76x cheaper and 5x faster in tests, after GPT-6 Astra in Codex ran a 144-run study across four models. The fixes were caching the growing browsing history, a larger history budget (480,000 characters) and removing screenshots in batches of 20. Average estimated cost fell to $0.47 per run on GPT-6.1 Sol, and all 18 runs answered correctly.

TL;DR Asana used Codex with GPT-6 Astra to test history caching and screenshot pruning, cutting browser agent cost 76x and runtime 5x.
Open original ↗

Organisations named here

  • OpenAI
  • Asana
  • Frank Hidalgo
  • StackAI

Technologies named here

  • Codex
  • GPT-6.1 Sol
  • GPT-6 Astra
  • Command
  • Model A
  • Model B
  • Model C
  • StackAI

Techniques named here

  • Prompt caching

Concepts named here

  • Cache-stable history
  • History budget

Related in agents