Fable 5.2 Is Out! It Built a Whole OS, It's 20% Cheaper & Why You Should NEVER Use Max
Claude Fable 5.2 is out. Anthropic's demo is an operating system it wrote from an empty folder with over a thousand subagents in eight days, the API price drops 20% to $8 in and $40 out per million tokens, and Anthropic says it's up to 30% faster. This video covers who gets it on each plan (Pro only has it in-plan through November 16), why the Max effort setting costs three times as much as Xhigh and scores lower, the benchmarks it loses, what people built in the first three hours, what breaks if you run agents on it, and the stronger unreleased model in the system card. 🔔 SUBSCRIBE TO BYTEFORWARD AI news, coding agents, model comparisons, benchmarks, and the things people are actually building: https://www.youtube.com/channel/UCGGHPOAzUaKsSc2yjsOeAFQ?sub_confirmation=1 CHAPTERS 0:00 Fable 5.2 is out 1:00 Who gets it on each plan 2:05 The rate card 3:05 Benchmarks, and where it loses 4:46 Effort levels and cost per task 6:49 What Anthropic built with it 8:17 What people built in three hours 10:22 What breaks if you run agents on it 11:17 Safeguards and Mythos 5.2 12:07 What Anthropic held back 13:24 What I'd do SOURCES Anthropic, Claude Fable 5.2 and Mythos 5.2 (launch post and system card): https://www.anthropic.com/claude-fable-and-mythos-5-2 Fable 5.2 launch offers (Pro week, limit resets): https://support.claude.com/en/articles/17506494-claude-fable-5-2-launch-offers Fable models on your plan: https://support.claude.com/en/articles/15424964-claude-fable-models-on-your-plan Claude plans: https://claude.com/pricing API pricing: https://platform.claude.com/docs/en/about-claude/pricing Fast mode: https://platform.claude.com/docs/en/build-with-claude/fast-mode What's new in Fable 5.2 (breaking changes, effort guidance): https://platform.claude.com/docs/en/models/fable-5-2/whats-new-fable-5-2 Claude Code changelog: https://code.claude.com/docs/en/changelog Artificial Analysis, Fable 5.2: https://artificialanalysis.ai/models/claude-fable-5-2 METR, Anthropic AI R&D acceleration assessment: https://metr.org/blog/2026-11-05-anthropic-ai-rd-acceleration-assessment/ Hacker News launch thread: https://news.ycombinator.com/item?id=50429906 Anthropic demos: https://www.youtube.com/watch?v=vVy0BRouX6E https://www.youtube.com/watch?v=vdYynH177h0 https://www.youtube.com/watch?v=54Oajk23wbI BUILDS AND POSTS @teortaxesTex (Minecraft in Three.js): https://x.com/teortaxesTex/status/2119899511317225567 @mattshumer_ (voxel Earth, early access): https://x.com/mattshumer_/status/2119877065701036047 @emollick (reverse Scooby Doo game, early access): https://x.com/emollick/status/2119876449494540554 @MatthewBerman (Unreal shooter, early access): https://x.com/MatthewBerman/status/2119871207206158459 @aimlapi (cost comparison, an API reseller's ad): https://x.com/aimlapi/status/2119889001893027840 @Yuchenj_UW (the effort chart): https://x.com/Yuchenj_UW/status/2119867942603264279 @theo: https://x.com/theo/status/2119868572386431281 @bcherny (limit resets): https://x.com/bcherny/status/2119873095129550880 @ArtificialAnlys: https://x.com/ArtificialAnlys/status/2119878995386982508 @claudeai: https://x.com/claudeai/status/2119867122683953298 Benchmark numbers are Anthropic's own unless the video says otherwise. The 200K context fix is from users and Anthropic hasn't confirmed it. The 5.5 name for the unreleased model is my guess. #Fable52 #Claude #Anthropic
