Asana cuts model costs 76x in browser tests with GPT-6.1 Sol
76x cheaper. Asana's browser agent now runs on GPT-6 Astra in Codex, slashing model costs while cutting latency 5x — a concrete win for enterprises deploying agents at scale.
Why it matters
Asana deployed GPT-6 Astra via OpenAI's Codex routing to cut browser-agent inference costs by 76x and latency by 5x in tests. This is a product efficiency story — cheaper, faster agents mean better margins and faster user experience for deployed systems, signaling how cost-optimized model routing is reshaping agent economics.
The key facts
6 to know76x cost reduction in browser-agent tests
5x latency reduction measured
Model: GPT-6 Astra via OpenAI Codex routing
Use case: Asana browser agent (production deployment context)
Status: Test results, not yet GA to all customers implied
Source: OpenAI official announcement (vendor claim, not independently verified)
Go to the source
OpenAI Blogopenai.com
Publisher excerpt: Using GPT-6 Astra in Codex, Asana made its browser agent 76x cheaper and 5x faster in tests to offer customers more capable models.