Video summary

I Tested Astra 6.2 vs. Fable 5.2. What You Need to Know.

Main summary

Key takeaways

Technology

Summary

The presenter compared GPT-6.2 Astra in Codex with Fable 5.2 in Claude Code across 12 practical tasks. Both models ran at high effort, received the same requests, and worked in separate branches. The presenter tracked API cost, token use, and runtime.

Results by task

Task Preferred model Main observations Website design Astra Both produced polished sites for the fictional Perk Form brand. Astra’s was cleaner and cost less; Fable’s had more animation and detail. Presentation Astra Both made professional, on-brand decks. Astra’s simpler slides suited presenting; Fable’s more varied, text-heavy design might work better as a document to read. Canva painting Astra Astra produced a more realistic portrait and worked around Canva’s drawing-stroke limit by exporting and continuing on a new canvas. Fable stopped at the limit. Event recap video Fable Both edits were judged poor and not energetic enough, but Fable’s was the better of the two. The presenter said Opus 5.5 had previously done better. 3D virtual summit Fable Fable’s world had working navigation, agendas, session playback, speakers, and sponsor booths, with an experience closer to the event brand. Astra looked more attractive but was less convincing as an experience. Instagram Reel Fable Both created reels from the same material, but Fable was preferred for sound design and B-roll selection. YouTube edit Fable Fable followed the intended editing workflow and style. Astra ignored the relevant skill and used a different tool, producing an unsatisfactory edit. Visual learning guide Astra Both turned a video into a tutorial with timestamps, screenshots, and annotations. Fable had more interactive structure, including progress tracking; Astra’s guide was considered tidier and much cheaper. Trip planning Fable Astra offered a better-looking but general itinerary. Fable provided more practical detail, links, and an interactive map, though its design was less polished. Instagram carousel Astra Neither model followed the established brand rules. Astra’s output was considered less cluttered, highlighting that skills and context may not transfer consistently between models. Financial workbook and dashboard Fable Both included formulas. Fable’s workbook had more detailed drivers, visualizations, and links between sheets; Astra’s was simpler and more visual. Browser game Fable Fable made the more engaging 3D game, with better physics and features such as a cat, flight, and checkpoints. Astra’s version felt slow and comparatively two-dimensional.

Final score: Fable 7, Astra 5. Fable won more tasks on output quality, but the presenter emphasized its substantially higher resource use. Astra was faster and cheaper in most comparisons; overall, Fable’s API cost was reported as 7.2 times higher, with substantially more tokens consumed. The video also notes that token efficiency does not guarantee better results.

The presenter’s broader takeaway was that Astra can perform well—especially when it uses the right tools and skills—but model behavior differs by task. Existing skills did not always work as expected across both systems, and better guidance or model-specific skills might improve results. The tests focused on practical creative and business tasks, not complex programming or large-codebase refactoring.

Guide and resource mentioned

  • The video demonstrates creating a visual tutorial from a YouTube video, with timestamps, annotated screenshots, playable clips, and—in Fable’s version—progress tracking.
  • The presenter briefly promotes Scrollcraft, a free web-design skill for scroll animations, layering, typography, spacing, and depth.

Main speaker/source

  • Nate Herk, presenting the comparison and reviewing the outputs.

Rate this summary

Your feedback will help improve summaries.

Improve this summary

Reprocess with a stronger model when the summary feels incomplete or inaccurate.

Pro

Translate summary in another language

Pro

Ask questions to this video

Chat for follow-up questions, clarifications, and source-backed answers.

Coming soon

Share this summary

Original video