{"slug": "claude-fable-5-2-one-day-later-video", "title": "Claude Fable 5.2, one day later [video]", "summary": "Henrik Kvamme reported on X that Anthropic's Claude Fable 5.2 tops benchmarks but costs $11,000 to $15,000 per build at API prices, and that usage in a $200 plan is worth about one fifth as much on Fable 5.2 as on Opus 5.5. On 30 ordinary coding tasks, two private tests found Fable 5.2 tied with Opus 5.5 while costing almost twice as much per solved task, with the gap appearing only on big, hard jobs. Kvamme produced a launch video for Fable 5.2 using 160 day-one user reports as the script, with research by six Sonnet 5.5 agents and script and animation by Fable 5.2; the video was not made or endorsed by Anthropic.", "body_md": "Henrik Kvamme on X: \"Fable 5.2 tops the benchmarks. So I had it make its own honest launch video, from 160 day-one user reports.\nHow to use it well:\n- Start with Opus 5.5\n- Give Fable the big, vague job\n- Set effort to high, not max\n- Don't let it grade its own work\nWhy, in the video:\" / X\n\nHenrik Kvamme on X: \"Fable 5.2 tops the benchmarks. So I had it make its own honest launch video, from 160 day-one user reports.\nHow to use it well:\n- Start with Opus 5.5\n- Give Fable the big, vague job\n- Set effort to high, not max\n- Don't let it grade its own work\nWhy, in the video:\"\n\nFable 5.2 tops the benchmarks. So I had it make its own honest launch video, from 160 day-one user reports.\nHow to use it well:\n- Start with Opus 5.5\n- Give Fable the big, vague job\n- Set effort to high, not max\n- Don't let it grade its own work\nWhy, in the video:\n\nFable 5.2 tops the benchmarks. So I had it make its own honest launch video, from 160 day-one user reports.\nHow to use it well:\n- Start with Opus 5.5\n- Give Fable the big, vague job\n- Set effort to high, not max\n- Don't let it grade its own work\nWhy, in the video:\n\n3/ The catch: at API prices, those builds cost $11,000 to $15,000 each.\nAnd the usage in a $200 plan is worth about one fifth as much on Fable 5.2 as on Opus 5.5.\n\n4/ On 30 ordinary coding tasks, two private tests found a tie with Opus 5.5. Fable cost almost twice as much per solved task.\nThe gap shows only on big, hard jobs.\n\n5/ Sources: game footage and cost estimates @MatthewBerman, plan math @theo, bug-hunt test @PawelHuryn, paper example @emollick, effort data @ArtificialAnlys.\nNot made or endorsed by Anthropic. Research: six Sonnet 5.5 agents. Script and animation: Fable 5.2.\n\nUsing real user reports as the script is a great idea, way more honest than the usual hype reel. As a motion designer making launch videos, I'm curious: how much did you art-direct the visuals vs. let it run?", "url": "https://wpnews.pro/news/claude-fable-5-2-one-day-later-video", "canonical_source": "https://twitter.com/xhenrikkvamme/status/2120137440169984242", "published_at": "2026-11-10 13:33:17+00:00", "updated_at": "2026-11-10 13:53:58.294566+00:00", "lang": "en", "topics": ["large-language-models", "artificial-intelligence", "ai-products"], "entities": ["Anthropic", "Claude Fable 5.2", "Claude Opus 5.5", "Henrik Kvamme", "Matthew Berman", "Theo", "Pawel Huryn", "Ethan Mollick"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/claude-fable-5-2-one-day-later-video", "markdown": "https://wpnews.pro/news/claude-fable-5-2-one-day-later-video.md", "text": "https://wpnews.pro/news/claude-fable-5-2-one-day-later-video.txt", "jsonld": "https://wpnews.pro/news/claude-fable-5-2-one-day-later-video.jsonld"}}