gemini-4IMAGE
Gemini 4 Argon (high) scores 20.2% on WeirdML
Gemini 4 Argon (high) scored 20.2% on WeirdML, landing between Sonnet 5.5 and Fable 5.1, strong but still behind the frontier on token efficiency.
Håvard Ihle
@htihleArchived 17 days agoOriginally posted 2026.10.18


Håvard Ihle @htihleOriginally posted 2026.10.18
Gemini 4 Argon (high) scores 20.2% on WeirdML, between Sonnet 5.5 and Fable 5.1. A strong performance, but still behind the frontier, especially on token efficiency.
This post quotes another post: x.com/htihle/status/2100892831187443782
x.com/htihle/status/2111782267345682840Archived 2026.10.19 10:00