Heidelberg AI News

Video · Cole Medin ·

Your Complete Guide to Fable 5.2 (When to Use It, Effort Levels, Prompting)

In short

The video explains when to use Fable 5.2 instead of Opus 5.5, how to split planning from implementation, and how effort levels and prompting affect results and cost. Fable is best for open-ended research and major decisions, while Opus is usually sufficient for clearly defined execution tasks.

Key points

  • Fable 5.2 costs about twice as much as Opus 5.5, reaching $26.54 per task at maximum effort versus less than half that for Opus.
  • Use Fable 5.2 for planning, research, and architectural decisions, then use Opus 5.5 to execute a documented plan.
  • Fable 5.2 generally performs best at high effort; maximum effort costs about $26.50 versus $9 at high effort with limited additional gains.
  • Opus 5.5 often works well at medium effort, achieving 92.8% on the cited benchmarks; high effort is sometimes appropriate.
  • Set cheaper models such as Sonnet or Opus for subagents instead of letting them inherit Fable 5.2.
  • Prompt Fable with the desired outcome, give it ownership of the codebase, and remove outdated rules from context files.

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

VideoRiley Brown

Claude Just Changed EVERYTHING (Here’s What’s New)

The video surveys new AI-agent capabilities across Claude, Gemini, GrokBot, and Dots, with emphasis on Claude’s workspace artifacts, meeting participation, model tradeoffs, and computer control. It argues that agents are becoming operational coworkers, but payment restrictions and platforms blocking agents remain major barriers to running businesses with them.

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.