Heidelberg AI News

Video · Riley Brown ·

Fable 5.2 is here

In short

The video livestream tests Claude Fable 5.2 through AI-assisted Blender video creation, presentation design, website redesign, and video-style reverse engineering. It highlights strong creative output and high-quality 4K rendering, but also slower execution, higher cost, and uncertain advantages over Opus 5.5 that require extended real-world testing.

Key points

  • Fable 5.2 generated a Blender video in one prompt and produced 4K-quality output.
  • Blender Cycles delivered realistic lighting, reflections, shadows, and depth of field, unlike faster Eevee rendering.
  • RunPod RTX 5090 GPUs accelerated rendering, with the final workload costing about $6.
  • Fable 5.2 was described as roughly twice as expensive as Opus 5.5.
  • The livestream compared Fable 5.2 with Opus 5.5 using benchmarks showing an estimated 4% to 8% advantage.
  • The model was also tested on presentation generation, website redesign, and reverse-engineering a Kurzgesagt-style video.
  • The creator emphasized that benchmark results are insufficient and that meaningful model evaluation may require about 40 hours of use.

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

VideoDavid Ondrej

My $1,000 Fable 5.2 Workflow (just copy me)

The video explains a coding workflow that combines Fable 5.2 for difficult planning and reasoning, Opus 5.5 for faster execution, and Gemini 4 Carbon for independent reviews. It matters because Fable is highly capable but consumes limits quickly, so model selection and task-based routing reduce cost and latency.