--- title: Anthropic and OpenAI both brag about app speed rewrites description: Anthropic and OpenAI both publicized speed improvements and rewrites of their flagship apps around the same time. Anthropic engineer bcherny teased a blog post... canonical: https://daily.dev/posts/anthropic-and-openai-both-brag-about-app-speed-rewrites-f0wdanhdx twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Anthropic and OpenAI both brag about app speed rewrites | daily.dev og:description: Anthropic and OpenAI both publicized speed improvements and rewrites of their flagship apps around the same time. Anthropic engineer bcherny teased a blog post... og:url: https://daily.dev/posts/anthropic-and-openai-both-brag-about-app-speed-rewrites-f0wdanhdx og:image: https://api.daily.dev/og/posts/F0WdanHdx.png og:image:alt: Anthropic and OpenAI both brag about app speed rewrites og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Anthropic and OpenAI both brag about app speed rewrites **[Trends](https://daily.dev/sources/trends)** · 2 min read · 2 upvotes · 1 comments ## Summary Anthropic and OpenAI both publicized speed improvements and rewrites of their flagship apps around the same time. Anthropic engineer bcherny teased a blog post on how Claude Code and the Desktop app got faster, framed as learnings for developers. OpenAI's ajambrosino announced ChatGPT Web was rewritten on top of the desktop app codebase in five weeks, amplified by retweets from Sam Altman and jxnlco. Both announcements are light on specifics—no benchmarks, no regression data, no cost details—leaving the actual engineering substance unverified and treating app polish as a competitive narrative rather than a proven outcome. ## Content Anthropic says Claude.ai and the Desktop app got roughly three times faster, and Theo (@theo) spent a video poking at the receipts. His verdict: the strategies are fascinating, and you can steal the best parts. He also found threads disappearing or failing to sync across browser sessions while live-testing the site. His read is that some of the speed came from aggressive caching that isn't invalidated enough, which trades latency for data staleness. The pitch comes from @bcherny, who pointed engineers to a blog post full of "juicy learnings & techniques." The setup: a two-week sprint where an internal Claude model, roughly Opus-class, worked autonomously in a Slack channel called Claude Tag. It proposed and shipped hundreds of PRs, with over 3,000 merged, behind human review and feature flags. The method is the part worth stealing. The team instrumented 13 measurements across four major user journeys. Instead of noisy wall-clock timing, they used deterministic proxies like V8 instruction counts and React commit counts. CI guardrails then ratchet the performance ceilings down over time, so regressions can't creep back in. Some of the fixes are great trivia: - Precompiling V8 code caches - Static HTML composers - Syntax highlighting moved to WASM and workers - A bug where em dashes forced UTF-16 string paths and slowed highlighting Theo compared this to his own T3 Chat and T3 Code work, which leans on Convex, local storage caching, and hover-based prefetching. Same goal, different bets. He also wandered into the old OpenAI/Anthropic engineering culture rift, and plugged sponsor WorkOS and its new agent sign-in standard (OMD). OpenAI is on a parallel track. @ajambrosino says the team spent 5 weeks completely rewriting ChatGPT Web on top of the desktop app codebase, and the post got boosted by @sama and @jxnlco. So both labs are now telling the same story: the chat app is getting rebuilt around the desktop codebase, with models doing a lot of the work. My take is that the speed numbers are probably real, and the cache bugs are the bill. Fast and wrong is still wrong, and "my thread vanished" is a worse failure than a slow spinner. ## Questions this post answers ### How did Anthropic make Claude.ai three times faster? Anthropic ran a two-week internal sprint where an autonomous Claude model (roughly Opus-class) proposed and shipped over 3,000 merged PRs in a Slack channel called Claude Tag, reviewed by humans behind feature flags. The team instrumented 13 measurements across four user journeys using deterministic proxies like V8 instruction counts and React commit counts, with CI guardrails that ratchet performance ceilings down over time. Specific fixes included precompiled V8 code caches, static HTML composers, and moving syntax highlighting to WASM and workers. _Teams chasing measurable frontend speedups can track real-world verdicts on tactics like these through daily.dev._ ### What problems did aggressive caching cause in Claude.ai after its speed improvements? A reviewer live-testing Claude.ai after Anthropic's speed-focused update found threads disappearing or failing to sync across browser sessions, attributing it to overly aggressive, under-invalidated caching that serves stale data in exchange for speed. The critique framed this as a classic trade-off: fast but wrong isn't actually a win, since cache invalidation bugs can silently corrupt the user's view of their own data. _Developers weighing cache aggressiveness against correctness can follow these kinds of real-world cache bug reports on daily.dev._ ### What did OpenAI do to ChatGPT Web around the same time as Anthropic's speed sprint? OpenAI completely rewrote ChatGPT Web on top of the desktop app codebase over a five-week period, according to a team member's public statement amplified by Sam Altman and others. No benchmark numbers accompanied the announcement, making it a rewrite claim rather than a measured performance comparison, in contrast to Anthropic's metrics-driven approach. _Anyone comparing how major AI products approach large rewrites versus incremental optimization can follow the back-and-forth on daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@tekn0wledg** · 0 upvotes > ngl, i've done 3 top to bottom re-writes of apps in the last 6 months with Claude as my copilot. They were all a tremendous success (cost reductions, infrastructure simplification, faster developer experience, less bugs). Makes sense to me, I've experienced it. ## Similar posts on daily.dev - [Opus 4.6 and Codex 5.3](https://daily.dev/posts/opus-4-6-and-codex-5-3-4ynvsyxn5) · Tech Lead Digest · 0 upvotes · 0 comments --- Tags: [#performance](https://daily.dev/tags/performance), [#openai](https://daily.dev/tags/openai), [#chatgpt](https://daily.dev/tags/chatgpt), [#anthropic](https://daily.dev/tags/anthropic), [#claude-code](https://daily.dev/tags/claude-code) [View this post on daily.dev](https://daily.dev/posts/anthropic-and-openai-both-brag-about-app-speed-rewrites-f0wdanhdx) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Anthropic and OpenAI both brag about app speed rewrites","url":"https://daily.dev/posts/anthropic-and-openai-both-brag-about-app-speed-rewrites-f0wdanhdx","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/anthropic-and-openai-both-brag-about-app-speed-rewrites-f0wdanhdx"},"datePublished":"2026-09-29T21:49:46.728Z","dateModified":"2026-10-05T21:00:02.888Z","description":"Anthropic and OpenAI both publicized speed improvements and rewrites of their flagship apps around the same time. Anthropic engineer bcherny teased a blog post...","isAccessibleForFree":true,"articleSection":"Trends","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":1,"discussionUrl":"https://daily.dev/posts/anthropic-and-openai-both-brag-about-app-speed-rewrites-f0wdanhdx","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"keywords":"performance,openai,chatgpt,anthropic,claude-code","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"Anthropic and OpenAI both brag about app speed rewrites"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/anthropic-and-openai-both-brag-about-app-speed-rewrites-f0wdanhdx","comment":[{"@type":"Comment","text":"ngl, i’ve done 3 top to bottom re-writes of apps in the last 6 months with Claude as my copilot. They were all a tremendous success (cost reductions, infrastructure simplification, faster developer experience, less bugs). Makes sense to me, I’ve experienced it.","datePublished":"2026-09-30T11:58:42.040Z","url":"https://daily.dev/posts/F0WdanHdx#c-d88higsN7","author":{"@type":"Person","name":"Scott Davis","url":"https://daily.dev/tekn0wledg","image":"https://avatars.githubusercontent.com/u/355212?v=4"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/anthropic-and-openai-both-brag-about-app-speed-rewrites-f0wdanhdx#faq","mainEntity":[{"@type":"Question","name":"How did Anthropic make Claude.ai three times faster?","acceptedAnswer":{"@type":"Answer","text":"Anthropic ran a two-week internal sprint where an autonomous Claude model (roughly Opus-class) proposed and shipped over 3,000 merged PRs in a Slack channel called Claude Tag, reviewed by humans behind feature flags. The team instrumented 13 measurements across four user journeys using deterministic proxies like V8 instruction counts and React commit counts, with CI guardrails that ratchet performance ceilings down over time. Specific fixes included precompiled V8 code caches, static HTML composers, and moving syntax highlighting to WASM and workers. Teams chasing measurable frontend speedups can track real-world verdicts on tactics like these through daily.dev."}},{"@type":"Question","name":"What problems did aggressive caching cause in Claude.ai after its speed improvements?","acceptedAnswer":{"@type":"Answer","text":"A reviewer live-testing Claude.ai after Anthropic's speed-focused update found threads disappearing or failing to sync across browser sessions, attributing it to overly aggressive, under-invalidated caching that serves stale data in exchange for speed. The critique framed this as a classic trade-off: fast but wrong isn't actually a win, since cache invalidation bugs can silently corrupt the user's view of their own data. Developers weighing cache aggressiveness against correctness can follow these kinds of real-world cache bug reports on daily.dev."}},{"@type":"Question","name":"What did OpenAI do to ChatGPT Web around the same time as Anthropic's speed sprint?","acceptedAnswer":{"@type":"Answer","text":"OpenAI completely rewrote ChatGPT Web on top of the desktop app codebase over a five-week period, according to a team member's public statement amplified by Sam Altman and others. No benchmark numbers accompanied the announcement, making it a rewrite claim rather than a measured performance comparison, in contrast to Anthropic's metrics-driven approach. Anyone comparing how major AI products approach large rewrites versus incremental optimization can follow the back-and-forth on daily.dev."}}]} ```