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

    Gemini 4 Argon tops Humanity's Sixth Sense benchmark with a 55.4% score

    Scale AI Labs says Gemini 4 Argon scored 55.4%, edging GPT-6-Astra's 53.6% on tests of reasoning beyond what's visible.

    1 Source, 1h ago, first seen 1h ago

    TLDR

    Scale AI Labs says Gemini 4 Argon leads its Humanity's Sixth Sense benchmark with a 55.4% score, ahead of GPT-6-Astra's 53.6%. The benchmark tests reasoning about hidden objects, what happens next and human behavior. Separately, a user recommends having Gemini critique screen-recorded tests of 3D games and interactive apps, saying it can catch flickering and audio discrepancies that screenshots may miss.

    Combined views

    —

    1 Source, first seen 1h ago

    — likes— comments— saves— reposts

    Combined views

    —

    1 Source, first seen 1h ago

    — likes— comments— saves— reposts
    Image: @ScaleAILabs on X

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Featured Source

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    1 Source

    Bilawal Sidhu@bilawalsidhuOpus and Astra are my daily drivers. But i love using both with Gemini for video reasoning tasks. ESPECIALLY useful if you’re making videos, 3d games or interactive apps that require actual play through to verify. Just ask your agents to screen record test candidates and have multimodal Gemini be the critic. Have them read GDMs agentic video understanding blog to learn how to prompt gemini for these tasks and turn it into a skill. It’ll catch tons of issues that are hard for agents to catch with screenshots alone eg. flickering, z-fighting, ui label collisions, audio discrepancies etc. Saves me a ton time and i wish I started doing this sooner.1h
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI

    1 Source

    Bilawal Sidhu@bilawalsidhuOpus and Astra are my daily drivers. But i love using both with Gemini for video reasoning tasks. ESPECIALLY useful if you’re making videos, 3d games or interactive apps that require actual play through to verify. Just ask your agents to screen record test candidates and have multimodal Gemini be the critic. Have them read GDMs agentic video understanding blog to learn how to prompt gemini for these tasks and turn it into a skill. It’ll catch tons of issues that are hard for agents to catch with screenshots alone eg. flickering, z-fighting, ui label collisions, audio discrepancies etc. Saves me a ton time and i wish I started doing this sooner.1h
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet