• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
  • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    Rohan Paul Claims GPT Astra Beat Claude Fable 5.1

    Post compares token counts and costs for generating HTML scenes from briefs.

    RP
    1 Source, 26d ago, first seen 26d ago

    TLDR

    Rohan Paul posted that GPT Astra outperformed Claude Fable 5.1 on an AI/ML API test. Each model received four briefs to convert into self-contained HTML scenes. Paul reported Astra used 25,159 tokens at $1.67 while Claude Fable 5.1 used 37,767 tokens at $2.51, a 33 percent reduction in both tokens and cost for Astra. The post includes a 28-second video showing side-by-side results. Paul is identified in the post as a Bengaluru-based machine learning engineer and Kaggle Master.

    Combined views

    18.9K

    1 Source, first seen 26d ago

    Combined views

    18.9K

    1 Source, first seen 26d ago

    109 likes
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    109 likes
    14 comments
    22 saves
    5 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    14 comments
    22 saves
    5 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    1 Source

    @rohanpaul_aiGPT Astra just beat Claude Fable 5.1 on this test by AI/ML API (@aimlapi ) . Astra used 33% fewer tokens than Claude Fable 5.1 and cost 33% less. GPT Astra: 25,159 tokens, $1.67 Claude Fable 5.1: 37,767 tokens, $2.51 GPT Astra and Claude Fable 5.1 were each asked to turn 4 briefs into self-contained HTML scenes. No external editing pass, one shot per scene. This kind of setup tests several things: interpreting the scene, deciding how to represent it procedurally, writing enough code to create the geometry and motion, and getting the whole program to work on the first attempt.

    1 Source

    @rohanpaul_aiGPT Astra just beat Claude Fable 5.1 on this test by AI/ML API (@aimlapi ) . Astra used 33% fewer tokens than Claude Fable 5.1 and cost 33% less. GPT Astra: 25,159 tokens, $1.67 Claude Fable 5.1: 37,767 tokens, $2.51 GPT Astra and Claude Fable 5.1 were each asked to turn 4 briefs into self-contained HTML scenes. No external editing pass, one shot per scene. This kind of setup tests several things: interpreting the scene, deciding how to represent it procedurally, writing enough code to create the geometry and motion, and getting the whole program to work on the first attempt.