Hype Check · Alex Finn

Worth your time?

Name: Is "Claude Fable 5 just dropped and I'm speechless..." worth watching? Hype 30/100 | Clip'd
Item: Claude Fable 5 just dropped and I'm speechless...
Rating: 70
Author: Clip'd

Claude Fable 5 just dropped and I'm speechless... by Alex Finn

Watch on YouTube

Hype score · lower is better

30/100

Useful / Authentic

Educational and practical with a live demo and clear how-to, plus a factual note about pricing and limited subscription access.

The 15-second version

Post it, send it, settle the argument.

Download clip

Receipts

Why this score

Video includes concrete CLI commands (/modclaw-fable-5, /slashgoal, /slashloop) and a live demo that produced a working app.
Alex gives practical settings (auto mode, slasheffort high) and shows outputs, not just marketing claims.
Pricing and access limitations are stated plainly without aggressive scarcity language or hard-sell.

Skip the video

The short version

Kept by Clip'd

Alex demos Anthropic's new Claude Fable 5 (aka Claude Mythos), showing how the model shifts users from micromanaging steps to giving high-level goals and running autonomous loops. He walks through setup in Claude Code (CLI/Ghosty), recommends settings (auto mode, high effort), demonstrates an 'advanced plan mode' to build a full personal productivity app in one loop, and notes short-term subscription access and pricing changes.

Keep the lesson

What it teaches

Treating an advanced LLM as an autonomous collaborator means you shift from micromanaging steps to setting clear goals plus success criteria, then verifying outcomes. Apply this model when the agent can access tools, run loops, and check its own work; avoid it for tasks requiring strict audit trails or legal compliance
Use 'advanced plan mode' as a way to force deep specification before execution: prompt the model to ask clarifying questions until you both share a full, testable spec. Use this whenever you need complex systems or multi-feature builds; skip it for tiny, one-off answers where overhead kills speed. The frequent error is

Jump straight there

Key moments

0:24Model claim: Alex shows benchmark results and raises skepticism about benchmarks versus real-world performance.
1:30Mental shift: Alex explains you no longer micro-instruct step-by-step, you now check whether Claude is doing the right work.
3:59Force-enable tip: Run /modclaw-fable-5 in the CLI to access Fable even if the model picker hasn't updated.
4:13Auto mode recommendation: Alex tells viewers to switch to auto mode (shift+tab) so the agent won't stop asking for permissions.
4:28Effort setting: He recommends starting with 'high' effort because Fable "torches tokens" with heavier thinking.
7:54Goal loop start: Use advanced plan mode so Claude asks numerous questions and builds a comprehensive plan before starting work.

Lines worth keeping

Quotes

It not only wins every single AI benchmark ever made, it dominates every AI benchmark ever made.0:24
This is quite literally super intelligence.2:14
Zero bugs. It just works. Works.10:38

Clip'd

Skip the video. Keep the lesson.

Score your next video free

The Hype Board