Heidelberg AI News

Video · Bijan Bowen ·

Claude Fable 5.2 Is INSANE – Is THIS Actually AGI?

In short

The video stress-tests Claude Fable 5.2 across coding, 3D game development, robotics, and physical electronics troubleshooting. It shows strong performance in C++ games, Blender/Godot scenes, and interactive PCB diagnosis, but less impressive 3JS visuals, high usage costs, and inconsistent results make the AGI claim premature.

Key points

  • Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, with Fast mode costing twice as much for 2.5x speed.
  • The Browser OS test ran for 1 hour 18 minutes and produced a creative but visually uneven set of mini-apps and games.
  • The C++ skate game took 1 hour 56 minutes and delivered strong physics, audio, environment detail, and gameplay.
  • The robot-arm test finished in 22 minutes, successfully grasped the car but dropped it off the table.
  • The Subway FPS test took 45 minutes and had solid gameplay responsiveness but only moderate visual quality.
  • The Blender/Godot pool game took 51.5 minutes and produced the strongest visual result, including reflections, lighting, and wet-dog behavior.
  • The PCB troubleshooting workflow gave interactive measurement guidance and correctly helped rule out a defective relay, demonstrating a practical physical-world use case.

Written by an AI model from the video’s captions and description.

Watch on YouTube The opening of the article at youtube.com.

In this story

A tool or model opens in the Heidelberg AI Curriculum: what it does, what it costs, how to start. A repo opens on GitHub.

Related stories

All stories

VideoWes Roth

Claude JUST got MYTHOS...

The video argues that Claude is evolving into an AI operating-system layer that coordinates office tools, meetings, data, software, and third-party apps rather than replacing each app directly. It also explains how eligible independent researchers, open-source developers, and vulnerability hunters may apply for Mythos 5.2 through Anthropic’s Cyber Verification Program, while challenging the interpretation of the MIT study claiming 95% of enterprise AI projects produce no ROI.