How we made claude.ai 3x faster in two weeks
Anthropic cut core claude.ai and desktop latency about 3x in two weeks, using Claude to find and ship fixes.
Anthropic engineers say they made the core experience of claude.ai and the Claude desktop app about three times faster during a two-week August sprint. At the 75th percentile, time to a typeable page on a fresh load fell from 3.1 seconds to 0.55 seconds, starting a Claude Code session fell from 0.8 to 0.3 seconds, and loading a Claude Cowork cloud session fell from 2.6 to 0.73 seconds. They focused on four journeys covering about 95% of activity and estimate the changes save tens of thousands of user-hours of waiting each day. Using Claude Tag with an internal model roughly comparable to Opus 5.5, the team merged more than 3,000 changes without a customer-facing incident or rollback.
- Fresh-load time to a typeable page fell from 3.1s to 0.55s at p75.
- New Claude Code sessions dropped from 0.8 seconds to 0.3.
- Cowork cloud session load fell from 2.6 seconds to 0.73.
- Team merged more than 3,000 changes with no customer rollback.
- Internal model near Opus 5.5 drove the work through Claude Tag.
- Vendors
- Anthropic
From the original article
15 min read at claude.dev
This August, we made the core user experience of claude. ai and the Claude desktop app about 3x faster in a two-week sprint. Users had been telling us it was slow, and they were right.
By matthieu_bl · claude.dev
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.