Skip navigation
Anthropic Managed Agents team:
0:00 / 13:23

Anthropic Managed Agents team:

Codez

@0xCodez51K followers

Visit on X
2,110 likesDownloadOriginal post

472,250 views

Anthropic Managed Agents team: "Fable 5 is our best model for running self-improving agent systems. Add /loops, dynamic workflows, dreaming and you are unstoppable" in 13-minutes, Anthropic team shows how to build self-improving agent systems with Fable 5 from scratch. Worth more than a $500 agent building course. Live from the last Anthropic stage in Japan. Unpublished.

40 Comments

Comments from the original post on X (Twitter)

  • Dexoryn's profile picture

    @Dexoryn

    Yes, I just experienced the new AI agent model : the Fable 5 to find the new Polymarket Trading strategy, and backtest. It works well, and Its analytics and recommendation was excellet. Thanks @0xCodez. Keep me your post.

    5 likes

  • Codez's profile picture

    @0xCodez

    you are a real legend Dexoryn ! Delivering best Polymarket agentic trading alpha. Love to follow you

    4 likes

  • Dexoryn's profile picture

    @Dexoryn

    Thanks

    1 like

  • Nikiton's profile picture

    @Nikitont

    Bro, I advise you to watch the latest broadcast @NVIDIAAI

    1 like

  • Codez's profile picture

    @0xCodez

    @NVIDIAAI What’s the most interesting thing you got here bro ?😅 Tease me

    1 like

  • Nikiton's profile picture

    @Nikitont

    @NVIDIAAI Check this out

    2 likes

  • Codez's profile picture

    @0xCodez

    @NVIDIAAI This looks like a solid watch. Thanks for sharing bro. Booking it

  • Nikiton's profile picture

    @Nikitont

    @NVIDIAAI Glad to be of help

    1 like

  • Chimpansky's profile picture

    @chimpansky

    what does 'dreaming' mean as a fable 5 feature specifically? is it the model running hypothetical rollouts internally, or a separate loop you configure?

    1 like

  • Harry Tandy's profile picture

    @HarryTandy

    saving $500 on a course is beautiful

    1 like

  • Codez's profile picture

    @0xCodez

    Building self-improving agent systems, even better bro 😆 Btw, have you tested Fable 5 already?

  • Alejandro Pereira's profile picture

    @aleGpereira

    Just one catch. You need millions of dollars

    1 like

  • Hussain Hashim | Building SundayBack's profile picture

    @itsthedonhashim

    @0xCodez whoa, this sounds intense. curious if they talk about real-world apps during the demo?

    1 like

  • Codez's profile picture

    @0xCodez

    Yes. They are talking about their clients cases too

    1 like

  • Hussain Hashim | Building SundayBack's profile picture

    @itsthedonhashim

    @0xCodez right? it's like every case has its own twists. do you have a favorite case type to work on?

    1 like

  • rewind's profile picture

    @rewind02

    loops are where fable wakes up

    1 like

  • Codez's profile picture

    @0xCodez

    Yeah. Was watching land recording live 😅

  • Alpha Batcher's profile picture

    @alphabatcher

    totally agreed with you Codez Fable 5 the best AI model ever

    1 like

  • Kekko D’Amato's profile picture

    @kekkodamato_

    The loop architecture is the real unlock here. Most people build single-shot agents, hit a wall on complex tasks, and blame the model. The key is adding an evaluation step *inside* the loop — not just at the end. The agent detects its own failures mid-run and retries with a different approach. That's actual self-improvement, not just repetition. Fable 5 maintains coherent state across those cycles in a way earlier models didn't.

  • Keysa's profile picture

    @SimplestBTCBook

    You seen this?